<%BANNER%>

Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.

Permanent Link: http://ufdc.ufl.edu/UFE0042084/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Klimas, Christie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Christie Klimas.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Kainer, Karen A.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042084:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042084/00001

Material Information

Title: Record for a UF thesis. Title & abstract won't display until thesis is accessible after 2012-12-31.
Physical Description: Book
Language: english
Creator: Klimas, Christie
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2010

Subjects

Subjects / Keywords: Forest Resources and Conservation -- Dissertations, Academic -- UF
Genre: Forest Resources and Conservation thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Statement of Responsibility: by Christie Klimas.
Thesis: Thesis (Ph.D.)--University of Florida, 2010.
Local: Adviser: Kainer, Karen A.
Electronic Access: INACCESSIBLE UNTIL 2012-12-31

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2010
System ID: UFE0042084:00001


This item has the following downloads:


Full Text

PAGE 1

1 MODELING COMPATIBILITY OF TIMBER AND NONTIMBER HARVEST S OF A MULTIPURPOSE AMAZONIAN SPECIES: ASSESSING SUSTAINABILITY THROUGH ECOLOGICAL AND ECONOMIC ANALYSES By CHRISTIE ANN KLIMAS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010

PAGE 2

2 2010 Christie Ann Klimas

PAGE 3

3 To Mark, Elizabeth and my parents: Thank you for your constant support and understanding

PAGE 4

4 ACKNOWLEDGMENTS I would first like to thank my advisor, Dr. Karen Kainer for her support and constant encouragement during my doctoral journey. I sincerely appreciate her thoughtful constructive feedback on everything from grant proposals to the various components of my dissertation. I also greatly appreciate her support, advice, and understanding as I balanced fieldwork with a new baby and her flexibility as I chose to write remotely. I would also like to thank Dr. Lcia Wadt for her assistance throughout my doctoral journey. In addition to providing logistical and institutional support for my research, she has opened her home to my family, helped me network with researchers throughout Brazil, and connected me to teaching and mentoring opportunities that have greatly enriched my doctoral research. I also sincerely appreciate the contributions from Dr. Christina Staudhammer and Dr. Wendell Cropper, who were instrumental in helping me work through analysis questions on my first two ar ticles as well as the feedback and suggestions of my other committee members Dr. Emilio Bruna and Dr. Doug Carter. I also want to thank my collaborators in Brazil. This dissertation was a collaborative international effort. My sincere thanks go to Valria RigamonteAzevedo and Manoel Freire Correia who coordinated phenology and seed production measurements while I was in the United States. I also want to thank the students and technicians who participated in this timeintensive data collection and improved the process through their valuable insights, including Llian Maria da Silva Lima, Antoninho Izidorio Petik, Aldeci da Silva Oliveira, Airton do Nascimento Farias, Pedro Raimundo Rodrigues de Arajo, Ana Claudia Costa da Silva and all those who provided p eriodic assistance during these years. I want to thank my friends and colleagues at Embrapa who have, and will hopefully continue to enrich my life and research. In particular, I want

PAGE 5

5 to thank both present and former directors of EmbrapaAcre, Dr. Judson F erreira Valentim and Marcus Vincio Neves DOliveira, for their support of my research. This project forms part of a larger project on nontimber forest projects. I want to thank project Kamukaia for their research support, especially researchers at Embrapa Roraima with I collaborate. I also want to thank Daniela Matias de Carvalho Bittencourt, Sabrina Gaspar, Andrea Raposo, Luciano Arruda Ribas, Paulo Wadt, Karina Martins, Sandra Tereza Teixeira, Sumaia Vasconcelos, Neuza Boufleuer, Cristina Lacerda, Elsa Mendoza, Monica de los Rios and Francisco Carlos for their friendship and support. I am grateful to the Rotary International Fellowship program for their initial support of my research. My sincere thanks go to Dr. Foster Brown who served as my research mentor during my initial year in Brazil and Francisco (Magnesio) and Marisa dos Santos who helped me adjust to life in Acre and introduced to me an incredible community of family and friends. I want to thank my Rotary host families, especially Maria das Graas, Jos Eduardo, Rodrigo, Ricardo and Roberto Moura Leite; Mariano, Socorro, Andr and Danielle Marques; and Marcos, Ana and Felipe Mendes. I want to thank my U niversity of F lorida (UF) friends and colleagues who have helped me on this journey, specific ally Joanna Tucker Lima, who spent hours providing useful feedback on various drafts of my dissertation chapters while in the midst of writing her own dissertation. I also want to thank my lab mates Marina Londres, Shoana Humphries, Cara Rockwell, Amy Duchelle, Rossa Cossio, Marlene Soriano, Vivian Zeidemann, and Jennifer Arnold, as well as Kelly Keefe, Ane Alencar, Hilary Zarin, Kelly Biedenwig, Mary Menton, Christine Lucas, Miramani Mishkin, Brian Daley, Jeff Luzar, Lucas Fortini, Amanda Holmes, Monica Morris, Amy Rosen and many others for their

PAGE 6

6 friendship, support and understanding during my graduate studies. I also want to thank all my friends on Facebook who provided virtual support and humor during the writing process (and to Amy Ripple and her secondgrade class who gave me an undeserved but appreciated honor by comparing me to The Lorax many thanks!). Since much of my writing was done out of state, I appreciate the welcome and support that I received from colleagues and friends in the Chicagoland area as well. I send my sincere thanks to Dr. Judith Bramble for her informal mentorship and support in my academic (and jobsearch) endeavors. I also want to thank Dr. Jim Montgomery for his assistance in networking and teaching. I appreciate the friendship and support of Dr. Sarah Richardson and Dr. Rick Hudson as well as their recommendation of Nihil Nimus, an excellent reference for academics at all career stages. I also want to thank Dr. Maggie Workman for her invitation to speak at a chemistry event which led to an adjunct teaching position. I also want to extend my sincere thanks to Dr. Allison Wilson for her belief in an untested instructor, and all my students at Benedictine University (fall and spring). I truly enjoyed teaching and learning with all of you. I thank you for providing me with the opportunity to expand my learning outside the area of my dissertation. I am continuously grateful to my family. My parents and siblings have supported my dreams, believed in me during my successes and struggles, and constantly pushed me to do my best irrespective of the obstacles. I also appreciate the constant questioning it keeps me on my toes. My parents have been instrumental in securing my excellent education and I have no doubt that much of my success (both past and present) stems from my education as well as their admirable parenting. I am also very

PAGE 7

7 lucky to have the support of my inlaws and incredibly grateful to have a husband who supported my research so much that he worked in Brazil for a time and has met many of my host families and friends. I appreciate his commitment to making my research work by absorbing extra childcare responsibilities in my absence or his willingness to spend time apart as Elizabeth and I travel in Brazil. Mark is a good voice of reason and experience pushing me to succeed in graduate school and afterward. I also want to thank my daughter, Elizabeth, for her incredible ability to see beauty in everything. Elizabeth has given me time to enjoy the spring flowers, play on the swings, build sandcastles, and run around the block. During preparation of this dissertation, Christie Klimas was supported by a College of Agricultural and Life Sciences Research Assistantship at the University of Florida (UF), a U.S. Environmental Protection Agency Science to Achieve Results (STAR) fellowship, UF Working Forests in the Tropics Integrative Graduate Education and Research Traineeship (IGERT), and a Tropical Conservation and Development Graduate Assistantship. The IGERT, a Woods Hole Fiel d Research Grant, Garden Club of America Caroline Thorn Kissel Summer Environmental Studies Scholarship, an Institute of Food and Agricultural Sciences travel grant, an Agricultural Womens Scholarship and the Explorers Club provided funds for research an d travel in Brazil.

PAGE 8

8 TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES .......................................................................................................... 11 LIST OF F IGURES ........................................................................................................ 13 LIST OF ABBREVIATIONS ........................................................................................... 14 ABSTRACT ................................................................................................................... 15 CHAPTER 1 INTRODUCTION .................................................................................................... 17 Modeling Variability in Seed Production .................................................................. 18 Ecological Viability of Combined Timber and NonTimber Harvests ....................... 19 Economic Revenue from Sustainable Harvest ........................................................ 20 2 MODELING VARIABILITY IN SEED PRODUCTION: A 5 YEAR STUDY TO EXPLORE THE INFLUENCES OF TREE ATTRIBUTES, HABITAT HETEROGENEITY, AND CLIMATE CUES ............................................................ 22 Introduction ............................................................................................................. 22 Methods .................................................................................................................. 25 Study Species .................................................................................................. 25 Study Site and Forest Types ............................................................................ 27 Tree Attributes and Climatic Variables ............................................................. 27 Phenology ........................................................................................................ 28 Below Tree Seed Counts ................................................................................. 29 Seed Removal and Infestation ......................................................................... 31 Data Analysis ................................................................................................... 32 Modeling variation in fruit production ......................................................... 32 Synchronicity in fruiting .............................................................................. 34 Results .................................................................................................................... 34 Phenology ........................................................................................................ 34 Seed Production ............................................................................................... 38 Seed Removal and Infestation ......................................................................... 44 Synchronicity in Fruiting ................................................................................... 44 Discussion .............................................................................................................. 44 Annual Patterns in Flowering and Fruitfall ........................................................ 45 Seed Production Variation by Forest Type ....................................................... 46 Influence of Forest Type Across Years ............................................................ 47 Variation by Tree Attributes .............................................................................. 49

PAGE 9

9 Influence of Climate Cues ................................................................................ 50 Management Considerations ............................................................................ 52 3 VIABILITY OF COMBINED TIMBER AND NON TIMBER HARVESTS FOR ONE SPECIES: AN EXAMPLE USING CARAPA GUIANENSIS ............................ 55 Introduction ............................................................................................................. 55 Methods .................................................................................................................. 59 Study Species .................................................................................................. 59 Study Site ......................................................................................................... 60 Individual Survival Rates and Diameter Growth ............................................... 61 Paramet er Estimation for Transition Matrices ................................................... 62 Matrix Modeling ................................................................................................ 66 Elasticity Calculations and Ranking .................................................................. 67 Modeling Environmental Stochasticity .............................................................. 68 Sustainable Harvest Simulations ...................................................................... 68 Quasi Extinction T hreshold Calculations .......................................................... 69 Results .................................................................................................................... 69 Vital Rates and Parameter Estimation .............................................................. 69 Matrix Modeling ................................................................................................ 70 Elasticity Calculations and Ranking .................................................................. 72 Modeling Environmental Stochasticity .............................................................. 80 Sustainable Harvest Simulations ...................................................................... 81 Discussion .............................................................................................................. 82 Sustainable Stochastic Harvest Es timates in Contrasting Forest Types .......... 82 Elasticity Calculations and Ranking .................................................................. 84 Limitations of the Harvesting Models ................................................................ 86 Implications for Multi Use Management ........................................................... 86 4 ECONOMIC REVENUE FROM SUSTAINABLE SEED AND TIMBER HARVESTS OF CARAPA GUIANENSIS ................................................................ 89 Introduction ............................................................................................................. 89 Methods .................................................................................................................. 93 Study Site and Field Measurements ................................................................. 93 Seed Oil Extraction and Price Estimates .......................................................... 94 Timber Value .................................................................................................... 95 Sustainable Harvest Simulations Under Simulated Environmental Stochasticity .................................................................................................. 96 Net Present Value of Seed and Timber Harvests ............................................. 98 Results .................................................................................................................... 99 Seed Oil Value ................................................................................................. 99 Timber Value .................................................................................................. 100 Net Present Value of Seed and Tim ber Harvests ........................................... 101 Discussion ............................................................................................................ 107 Shadow Prices ............................................................................................... 107 Model Assumpti ons ........................................................................................ 108

PAGE 10

10 Limitations of Model Results ........................................................................... 109 Best Management Practices for MultipleUse C. guianensis Management .... 110 5 CONCLUSION ...................................................................................................... 113 Research Significance .......................................................................................... 116 Evolution of Research and Collaborat ion Interests ............................................... 117 Future Research ................................................................................................... 119 LIST OF REFERENCES ............................................................................................. 120 BIOGRAPHICAL SKETCH .......................................................................................... 140

PAGE 11

11 LIST OF TABLES Table page 2 1 The number of Carapa guianensis trees monitored for estimates of seed production each year and t he time period in which production was quantified. .. 29 2 2 Results explaining variation in seed production, calculated using PROC GLIMMIX to perform a Poisson regression ......................................................... 39 2 3 Results of the model explaining variation in seed production when only 5week peak seedfall periods were used for all years. (For brevity, estimated parameters for year and year x forest type interactions are not shown) ............. 41 2 4 The average number of seeds per hectare ( _x ) and standard error (SE) for Carapa guianensis in kilograms. ......................................................................... 42 2 5 Number of individuals sampled and average seed weight per tree ( _x ) and standard error (SE) for Carapa guianensis .. ....................................................... 48 3 1 Number of trees wit h dendrometer bands installed in 2003 and measured in 2004. N reflects original tree numbers sampled in 2004 prior to subsequent tree mortality. ...................................................................................................... 61 3 2 Leftkovitch, or sizedependent, demogr aphic matrices for Carapa guianensis in two forest types: occasionally inundated and terra firme For each forest type, there are four annual transition matrices. ................................................... 73 3 3 Population asymptotic growth rates ( ) calculated from the Leftkovitch model for populations of Carapa guianensis in Acre, Brazil. ......................................... 75 3 4 Ranking of top four vital rate elasticity values in order of importance and their noted in parentheses. ......................................................................................... 79 3 5 Simulated stochastic population growth rate ( s) with simulated 95% confidence intervals and stochastic growth as calculated by Tuljapurkar's T) for populations of Carapa guianensis in Acre, Brazil.. .......... 80 4 1 Expected revenue per hectare of C. guianensis seed and timber harvest in an occasionally inundated forest (A) and terra firme forest (B) in Acre, Brazil. All revenues indicate the total revenue over 51years or 3 timber cutting cycles (mean, minimum and maximum values are from 500 simulations). ....... 102

PAGE 12

12 4 2 Expected revenue per hectare of C. guianensis seed and timber harvest in an occasionally inundated and terra firme forest in Acre, Brazil. All revenues indicate the total revenue over 51y ears or 3 timber cutting cycles (mean, minimum and maximum values are from 500 simulations). .............................. 104 4 3 Average number of individuals harvested per hectare for timber based on a 25year cutting cycle. ........................................................................................ 106

PAGE 13

13 LIST OF FIGURES Figure page 2 1 Percentage of individuals with new leaves (top panel) flowering (middle panel) and with observed fruitf all (bottom panel) in terra firme and occasionally inundated forests superimposed on monthly precipitation values (mm). .................................................................................................................. 35 2 2 Average number of Carapa guianensis seeds per tree. Estimates include measured seed production data plus an estimate of nonmeasured seed production for the first three study years, when seed production was quantified only during a 5 to 17week peak period. ............................................ 39 2 3 Relationship between d.b.h. and seed production using least squared mean values. ................................................................................................................ 41 2 4 Relationship between year and seed production using least squared mean values. ................................................................................................................ 43 3 1 The life cycle of C. guianensis with straight arrows indicating the probability that an individual will grow (Gi) from one stage to the next during time interval t Curved arrows indicate the probability that an individual will survive and remain in a given size class (Pi) during time interval t ....................................... 63 3 2 Matrix entry calculations ..................................................................................... 64 3 3 Annual sizeclass dependent diameter growth (mean + SD) of C. guianensis in terra firme forest (a) and occasionally inundated forest (b). Years included are 20052006, 20062007, 20072008 and 20082009. .................................... 71 3 4 Elasticity values for Carapa guianensis vital rates in two forest types: occasionally inundated (A, C, E) and terra firme (B, D, F). For each forest type and vital rate, y axis values vary and there are four elasticity bar s, one for each transition matrix. ................................................................................. 76 4 1 Potential annual seed oil production per hectare with harvests of 10% and 20% of the seed crop in occasionally inundated (top panel) and terra firm e forest (lower panel). Variability is due to modeled variability in seed production between years. ............................................................................... 105 4 2 Histograms of the distributions of NPVs for 10% seed harvest predicted from 500 sim ulations using a market price of R$25 and a labor cost of R$12; variability is due to differences in expected seed production based on modeled ecological stochasticity. ..................................................................... 106

PAGE 14

14 LIST OF ABBREVIATION S AIMEX Association of Timber Industry Exporters in the State of Par CV Coefficient of Variation CVp Coefficient of Variation (population level) DBH Diameter at Breast Height EMBRAPA The Brazilian Agricultural Research Corporation INMET Brazilian National Institute of Meteorology IP GRI International Plant Genetic Research Institute Population Growth Rate s Stochastic Population Growth Rate T Population Growth Rate calculated by Tuljapurkars Approximation MTF Multiple use Forest Management NPV Net Present Value NTFPs Non timber Forest Products PCCs Pearson Correlation Coefficient s Pi Probability of Survival PV Present Value RIL Reduced Impact Logging SE Standard Error UFAC Federal University of Acre xCVi Coefficient of Variation (between individuals)

PAGE 15

15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MODELING COMPATIBILITY OF TIMBER AND NONTIMBER HARVEST S OF A MULTIPURPOSE AMAZONIAN SPECIES: ASSESSING SUSTAINABILITY THROUGH ECOLOGICAL AND ECONOMIC ANALYSE S By Christie Ann Klimas December 2010 Chair: Karen A. Kainer Major: Forest Resources and Conservation This dissertation examines the relative compatibility of timber vs. nontimber forest product ( NTFP ) harvest of Carapa guianensis for po pulation persistence and income generation in two forest types in the Brazilian Amazon. C. guianensis is a multi use tropical tree valued for its seed oil and its highquality timber. This dissertation focused on variable components of sustainable harvest, including quantifying patterns of na tural seed production, modeling sustainable seed and timber harvests based on underlying demographic data collection, and simulating economic returns from seed and timber harvest s by calculating the equal annual equival ent (EAE) of revenues associated with mo deled simulations of seed and timber harvests. Our study demonstrated that multiple variables interact at different scales to influence seed production. Climatic cues were central to setting overall patterns in phenophases and seemed to best explain why high seed production years were consistent across both fores t types examined. I ndividual tree attributes contributed to seed production heterogeneity within habitats; dbh, and crosssectional canopy area showed posit ive, quadratic relationships with seed production, while vine load negatively affected seed production, irrespective of forest

PAGE 16

16 type. In our upland forest sites, demographic parameters suggested that no tested levels of seed and timber extraction were sustainable; quasi extinction simulations indicated population decline regardless of harvest levels. R emoving 10% of the total seed production proved feasible and consistent with a stable or increasing population in occasionally inundated forests though the q uasi extinction threshold of less than 3 individuals dbh 10 cm per hectare was reached within the 500year time interval. We also found the potential for compatibility between seed and timber harvest in occasionally inundated forests. Our results suggested that 10% annual seed harvests were compatible with 100% timber harvests (of trees equivalent of approximately two trees per hectare. R evenue from 100% timber harvests (of individuals years) exceeded that of 10% seed harvests when evaluated separately and applying labor costs of R$23 per kilogr am of oil (US$12.88). When we dropped those labor cos ts to R$12 (US$6.72), however, seed harvest EAE surpassed that of timber harvest In practice, however, benefits could accrue from both annual seed harvest revenues and per iodic logging of C. guianensis individuals. Findings add to the ecological understanding of the mechanisms contributing to variable yearly seed production and represent an important first step in providing information geared toward encouraging ecological b est m anagement practices while maintaining profitable harvest scenarios.

PAGE 17

17 CHAPTER 1 INTRODUCTION This dissertation examines the relative compatibility of timber vs. nontimber forest product ( NTFP ) harvest of Carapa guianensis for population persistence and income generation in two forest types in the Brazilian Amazon. The central dependent variables in this study are seed oil, timber production and price/cost variables. As such, each chapter of my dissertation addresses one of these variables. This dissert ation is structured so that the second, third and fourth chapters are independent articles ready for submission to peer reviewed journals. Relevant conclusions are found at the end of these three chapters and I articulate overall conclusions in the final c hapter (Chapter 5). Carapa guianensis Aublet. (Meliaceae) common name andiroba, is a multiple use species. As with other species in the Meliaceae family (Swietenia macrophylla and Cedrela odorata), C. guianensis is an important timber tree in the Neotropics (McHargue & Hartshorn 1983, Mabberley 1987, Dayanandan et al. 1999). It is also valued for the high quality oil extracted from its seeds (Shanley 2005). Pure Carapa seed oil is used for medicinal applications (Rodrigues 1989), with valueadded product s including soaps, shampoos, candles and repellent torches (Shanley 2005). The oil also enjoys international demand: between 1974 and 1985, Brazil exported between 200 300 tons of oil annually (Clay et al. 2000). This species is considered to have such g reat economic potential that the Amazonian State of Acre in Brazil has identified it as one of six priority species for extraction research (Acre 2000). As a valuable timber and nontimber forest product (NTFP), exploitation of C. guianensis promises to ex pand as market demand for seed oil and timber advances concomitantly int o the interior of the Amazon with federally funded highway development projects (Fearnside 2005).

PAGE 18

18 Modeling Variability in Seed P roduction C. guianensis management must take into account species value and productivity as both a timber and nontimber resource. Chapter 2, therefore, specifically addresses the NTFP resources and seeks to illuminate natural patterns of C. guianensis seed production and variability and how these patterns are related to underlying habitat heterogeneity, tree attributes and climatic cues. Unsustainable resource extraction can stem fro m a lack of understanding of species biology (Peters 1996). This is further complicated in cases where the resource of interest i s tree fruits or seeds: most polycarpic woody plants exhibit variable years of high and low seed production at the populationlevel (Herrera et al. 1998). In chapter 2, we tested the hypothesis that seed production is influenced by multiple interacting fac tors across scales (climatic cues, habitat heterogeneity, and tree attributes). To do this, we recorded reproductive phenophases of C. guianensis and quantified C. guianensis seed production over 5 years, using a Poisson regression to see how seed product ion was related to predictor variables at each scale. We predicted that: 1) Climatic cues, like rainfall, would be most important in determining populationlevel seed production due to potential for largescale resource limitation; 2) Seed production would be significantly higher in occasionally inundated forests based on observations by forest residents and previous research that documented higher C. guianensis densities in swamp forests; and 3) Tree attributes such as dbh and vine load would be strongly i mplicated in determining seed production potential. Findings from this study add to the ecological understanding of the mechanisms driving variable yearly seed production, particularly in tropical trees, while also contributing species specific information for C. guianensis m anagement.

PAGE 19

19 Ecological Viability of Combined Timber and NonT imber Harvests Determining ecologically sustainable harvest limits is necessary for developing appropriate forest management strategies, though effective simulation modeling o f tropical forests managed for NTFP and timber is a significant challenge (Ribeiro do Valle et al. 2007). We used information on variable seed production along with other demographic data to parameterize a matrix model developed to examine sustainable harv ests levels of both timber and seeds for this species (chapter 3). We question whether management of Carapa guianensis a tree species valued for both its high quality timber and its seed oil, can be managed sustainably for both seed and timber harvests. M atrix models are powerful tools for identifying population trends and responses to management regimes (Leftkovitch 1967, Crowder et. al. 1994, Olmsted & Alvarez Buylla 1995), setting harvest limits from a population that are compatible with its continued existence (Nantel et al. 1996, Ratsirarson et al. 1996) and identifying key life stages as management targets (Crouse et al. 1987). In addition, matrix models can be used to simulate sustainable harvest levels for timber, NTFPs or a combination of managem ent strategies for single species, an important factor given that one third to alm ost half of timber species also have NTFP value (Martini et al. 1994, Ndoye & Tieguhong 2004, HerreroJuregui et al. 2009). Our objectives were: 1) to quantitatively assess the impact of seed and timber harvesting on C. guianensis in stochastically varying environments, 2) to determine whether the impact of harvest varied between occasionally inundated and upland ( terra firme ) forests, and 3) to simulate potential ecologicall y viable harvest scenarios of C. guianensis seeds, timber, or seeds and timber. We hypothesized that intensive seed collection would still result in sufficient seedling recruitment to ensure continued species persistence, and that terra firme

PAGE 20

20 forests would allow even more intensive seed harvests as seeds in this environment may be safesite limited. Therefore, collection of seeds that would otherwise perish would allow for increased seed harvest levels without impacting future population viability. Econom ic Revenue from Sustainable H arvest Quantifying revenue from harvest is particularly important in cases where an individual species provides multiple economic benefits, generating a potential conflict of interest in deciding whether to prioritize one use over another. Such conflict of interest is common in the Amazon. Herrero Juregui et al. (2009) found that 46% of the 200 timber speci es in the state of Par, Brazil also provided nontimber forest products. Multiple use species should be managed to maxi mize their economic or social value under constraints of ecological sustainability. The determination of how to do this is often left to the forest resident or manager, using imperfect and general information on the benefits of different forest management strategies (Macpherson 2007). Availing quality information to managers should improve both ecological and economic decisionmaking, though detailed knowledge of sustainability does not always translate into sustainable management practices (Gaoue & Ticktin 2009) and expected economic gains are not always realized (Putz et al. 2000). Therefore, this research (chapter 4) was conducted, partly, to enhance the economic information available to the forest manager, in particular the potential revenue associated with different harvesting strategies of Carapa guianensis We modeled the equal annual equivalent (EAE), a function of net present value (NPV), of revenues associated with modeled simulations of sustainable seed and timber harvests. Our specific objectives were 1) to simulate and compare the revenue from ecologically viable seed and timber harvests of C.

PAGE 21

21 guianensis in stochastically varying environments; and 2) to simulate the EAE of revenues from seed and timber harvest under different market prices for bo th seed oil and timber. Sustainable harvest simulations are based on a stochastic ecological matrix model parameterized with data from a demographic study in the southwestern Amazonian state of Acre, Brazil (See Chapter 3).

PAGE 22

22 CHAPTER 2 MODELING VARIABILITY IN SEED PRODUCTION: A 5 YEAR STUDY TO EXPLOR E THE INFLUENCES OF TREE ATTRIBUTES, HABIT AT HETEROGENEITY, AN D CLIMATE CUES Introduction Seasonal patterns in tropical plant phenology are assumed to represent adaptive responses to proximate cues (De Steven e t al. 1987), though these cues are still unknown for the majority of tropical plants ( Stevenson et al. 2008) and can vary significantly by species (Rivera et al. 2002, Diaz & Granadillo 2005, Sing & Kushawaha 2006) Understanding the mechanisms that underl ie the seasonal phenology of individual tropical species and communities is increasingly important in efforts to understand the potential impacts of climate change on tropical plants ( Cleland et al. 2007, Stevenson et al. 2008), how plant populations may be affected by more frequent wea ther abnormalities like droughts ( IPCC 2007, Cox et al. 2008), and even phenologys adaptive evolution (Elzinga et al. 2007). Quantifying the relevant cues underlying plant phenology of species may indicate those species most vulnerable to expected abiotic changes For species that are important sourc es of marketable resources, ecological vulnerability may translate into an economic and social vulnerability for persons depending on resource revenue, an already vulnerable commu nity ( Rodrigues et al. 2009) A better understanding of the underlying mechanisms of reproductive phenology can improve species management options, especially for species whose seeds are a marketable resource T emperature and photoperiod are likely the best studied climatic cu es influencing tropical phenology (Elzinga et al. 2007, Stevenson et al. 2008) and have also been hypothesized to affect subsequent seed output (Kelly and Sork 2002). Climatic variables

PAGE 23

23 are clear cues for some species (Kelly et al. 2000, Schaumber et al. 2002, Monks & Kelly 2006), but remain inconsistent for others (Sork & Bramble 1993, Piovesan & Adams 2001). While much research on climatic cues that influence variable seed production has focused on temperate tree species, there is ev idence that temperature and rainfall also influence flowering of tropical plants ( Bernier 1988, Daz & Granadillo 2005) Low minimum temperatures signal bud development in peninsular Malaysia and Sarawak ( Ashton et al. 1988). Several studies have shown that rainfall patterns are responsible for seasonal cycles of growth, and flowering of some species in tropical environments (Borchert et al. 2004, McLaren & McDonald 2005, Brearly et al. 2007). Dry season severity is another factor, as many tropical forest plants require a threshold level of drought before successful flowering ( Alvim 1960; reviewed by Kozlowski & Pallardy 2002). Climatic variation is often captured by year to year quantification of the number of individuals flowering and subsequent seed production; most polycarpic woody plants exhibit variable years of high and low seed production at the populationlevel (Herrera et al. 1998). Distributions and abundances of many species are influenced by local heterogeneity in physical habitat variables (Webb & Peart 2000, Svenning et al. 2006), including soil nutrients (Gartlan et al. 1986, Baillie et al. 1987), drainage (Newbery et al. 1986, Hubbell & Foster 1986), soil texture (Davies et al. 1998) light availability (Lieberman et al. 1995) and topography ( Poore 1968, Ashton 1976, reviewed in Whitmore 1984). The importance of habitat heterogeneity in representing resource availability for plant growth, maintenance and reproduction is reflected in the

PAGE 24

24 predominance of classifications such as forest type a term we use to distinguish localized heterogeneity in our study site. While the interaction between habitat and climatic factors may influence phenology and seed production at the populationlevel, individual characteristics are also important in determi ning the flowering and subsequent seed production of specific trees. Since plants accumulate resources over time, size, as expressed by tree diameter, is positively correlated with seed production ( Ares & Brauer 2004, Snook et al. 2005, Kainer et al. 2007) Other associated factors, such as liana loads in crowns, have been found to reduce fruit production (Stevens 1987, Kainer et al. 2006) and suppress tree growth ( Clark & Clark 1990, Grauel & Putz 2004). There is also evidence that some trees are consistently high producers (Snook et al. 2005), perhaps due to correlations with larger crown size (Kainer et al. 2007) or the underlying genotypic characteristics of individuals ( Vander Wall 2001). This study tests the hypothesis that seed production is influence d by multiple interacting factors across scales ( climatic cues, habitat heterogeneity, and t ree attributes ) To do this we first recorded reproductive phenophases of C. guianensis and quantified C. guianensis seed production over 5 years using a Poisson regression to see how seed production was related to predictor variables at each scale. We chose C. guianensis as a model species because it is an economically important tropical tree that demonstrates pronounced temporal variability in seed production. W e predicted that : 1) Climatic cues, like rainfall, would be most important in determining populationlevel seed production due to potential for largescale resource limitation; 2) Seed production would be significantly higher in occasionally inundated forests based on

PAGE 25

25 observations by forest residents and previous research indicating that the species is found in higher densities in swamp forests (McHargue & Hartshorn 1983); 3) Tree attributes such as dbh and vine load would be strongly implicated in determining seed production potential. In addition to adding to the ecological understanding of the mechanisms contributing to variable yearly seed production, findings from this study also contribute to species management: C. guianensis is a tropical tree valued for its high quality timber and seed oil (Shanley & Medina 2005). Method s Study S pecies C. guianensis is a shade tolerant tree found in both canopy and sub canopy levels in a variety of tropical forest types. Mature C. guianensis trees can reach heights o f 30 meters (Ferraz et al. 2002) and diameters of 2 meters (Fournier 2003). Tree size may not reliably predict age; carbondating of two individuals in a nearby forest in the southwestern Amazon estimated the age of a 37.5 dbh individual at 172 years while a 17 cm dbh individual was aged at 785 years (Vieira et al. 2005). This species exhibits prolonged synchronous flowering (Hall et al. 1994), although the timing and length of this phenophase varies by region (Leite 1997). Pollinators include stingless be es (Apidae) and butterflies (Riodinidae and Lycaenidae) (Maus 2006). Seeds are enclosed in a dehiscent capsule that normally breaks open upon impact with the ground, freeing seeds for germination or dispersal via water or frugivores (McHargue & Hartshorn 1983a, Plowden 2004). Seeds are subject to attack by Hypsipyla ferrealis moth larvae. H. ferrealis larvae eat the endosperm, but in some cases dont consume enough to prevent seed germination or cause seedling mortality.

PAGE 26

26 Partial endosperm consumption can r esult in smaller seedlings due to diminished seed reserves (McHargue & Hartshorn 1983b). Some authors have published rough estimates of seed production ranging from 50 to 200 Kg of seeds per tree (Shanley et al. 1998, Shanley 2005). Other published seed p roduction estimates are based on samples of fewer than 10 trees (Ranklin 1978, McHargue & Hartshorn 1983a Plowden 2004, Guedes et al. 2008), though more comprehensive surveys of production were recently conducted in Amazonian floodplain forests (Londres 2009) and in the northwestern Brazilian state of Roraima (Tonini et al. 2008). Annual variability in seed production of this species, however, is still poorly understood, including the proportion of trees that suspend production over multiple years, an important factor in calculating overall yield and potential revenue. There may also be local variation in seed production due to differences in forest types: C. guianensis is preferentially a species of periodically inundated swamp forests (McHargue & Hartshor n 1983), though more recent research has shown high densities in a variety of environments, including terra firme and occasionally inundated forest (Klimas et al. 2007, Londres 2009). C. guianensis seeds reportedly provide food for many species of peccar ies, rats and large rodents (McHargue & Hartshorn 1983b, Fournier 2003). Both field observations and conversations with local hunters, however, indicate that C. guianensis seeds are not a favorite food for game animals near our study site. Using motionact ivated c ameras in highly altered forests, Silva (2009) found that rats ( Proechimys spp.) and acouchis ( Myoprocta pratti ) were the most common visitors to sites with C. guianensis seeds. These animals made only rare visit s during the dry season, when

PAGE 27

27 other food is scarce (Silva, 2009) On the other hand, faunal interest in seeds may be somewhat higher in less disturbed forests. Study Site and Forest T ypes The study was carried out in the 1200 ha forest reserve of the Brazilian Agricultural Research Corporat ion (EMBRAPA) (latitude 10 01' 28" S and longitude 67 42' 19" W), in the southwestern portion of the State of Acre, Brazil. Average annual temperature is 24.5C with a dry season from June to August (INMET 2009). The region has slightly hilly topography including upland habitat ( terra firme ) and occasionally inundated areas. This region is close to the southwestern limit of the species range and these forests differ from the floodplain forests where most C. guianensis research has been conducted (Guariguata et al. 2002, Plowden 2004, Londres 2009). We established four 400 x 400 m (16ha) plots within the reserve: two in areas of predominantly terra firme forest, and two in occasionally inundated forest. Tree Attributes and Climatic V ariables For each C. guianensis > 10 cm dbh in the four plots, we recorded spatial coordinates, canopy position [(dominant, codominant, intermediate or suppressed) from Smith et al. (1997)], diameter at breast height (dbh), tree height, crown cross sectional area, vineloading, tree abnormalities, and reproductive status. The latter three variables were assigned binary values. Trees with > 75% of their crowns loaded with vines were given a value of 1, whereas trees with <75% were assigned a value of 0. Tree abnormalities were defined as any characteristic that differed from most trees and could negatively affect tree function, such as evidence of a previous burn, forked trunks (a potential indication of prior insect attack), and damaged or absent crowns. Reproductive status w as determined based on evidence of seed production either on or

PAGE 28

28 below the tree. Crown cross sectional area was measured along two axes: maximum crown diameter and a second diameter at a right angle to maximum diameter. An ellipsoid formula was used to conv ert axis diameters to crown area (Kainer et al. 2007). Total tree height was measured with a SUUNTO optical height meter PM 5/1520 (clinometer), using a surveyors tape to determine distance from clinometer to tree. Rainfall and temperature data were coll ected daily at noon at the Federal University of Acre (UFAC) approximately 8 km from our study site. Monthly mean temperature was averaged from the daily compensated mean temperatures for a given month and maximum and minimum temperatures also represented monthly averages of these daily values (INMET 2009). Phenology We monitored the phenophases of C. guianensis from September 2005 to June 2009 for a random sample of trees in each of the two forest types, stratified by diameter class. We selected this sam ple from the larger sample of all inventoried individuals in three of the plots via random sampling ; one plot was excluded due to time constraints involved with weekly seed collection Trees with crowns overlapping with other conspecifics and those whose s eeds fell into adjacent rivers were excluded from the phenology sample and replaced by another randomly selected individual that met the criteria. Initially, we observed 53 trees located in occasionally inundated forest and 23 in terra firme After the fir st year, we added 28 additional randomly selected trees in terra firme forest, for a total of 51 trees. Diameter classes were 15 cm < dbh ranged from 7 to 18 tre es depending on natural density and probability of reproduction (Klimas et al. 2007).

PAGE 29

29 We used binoculars to conduct crown observations (OBrien et al. 2008) twice a month, beginning in September 2005. We recorded presence/absence of new leaves, flowers and falling fruits. We also checked the ground below canopies, looking for additional evidence of phenophases, such as the presence of fallen flowers. If we observed flowers either on the ground below the canopy or in the canopy, the individual was classified as flowering. C. guianensis flowers are small and ground observations served both to validate binocular observations and identify flowering trees where flowering was limited and canopy flowers were hidden from view. Once three or more individuals from t he entire sample population initiated flowering, we intensified monitoring from twice a month to weekly, which continued until the end of fruitfall. Based on observations of flowering and fruitfall outside of peak periods and rapid progression through phen ophases, we monitored phenology weekly from September 2007 until June 2009 (Table 2 1). Table 2 1 The number of Carapa guianensis trees monitored for estimates of seed production each year and the time period in which production was quantified. Quantificat ion period Measurement period (weeks) Number of trees Occasionally inundated Terra firme Nov / 2004 Mar / 2005 17 19 20 Jan / 200 5 Feb/ 2006 5 30 24 Jan / 200 6 Feb/ 200 7 5 29 23 2007 2008 C ontinuous 53 5 1 2008 2009 C ontinuous 53 50 Below Tree Seed C ounts We quantified weekly seed production by collecting fallen seeds under a random subset of trees selected from the phenology study population. The measurement period

PAGE 30

30 and number of sample trees in each environment varied annually. For our study annual seed production refers to the ecological seed production period for this species which can extend from November of one year to June of the following year in this region. In 20042005, we collected seed production weekly throughout a 17week peak see dfall period. In 20052006 and 20062007, we collected weekly seed production during 5week peak seedfall periods (Table 2 1). From September 2007 to July 2009, we collected seeds at weekly intervals throughout the year to account for seed production out side of peak periods, and measured production for each of the 104 trees in the phenology dataset (Table 2 1). Sampled trees were visited weekly on the same day or within a twoday period if rain or inclement weather delayed seed collection. Within 24 hours of collection, seeds were taken to an onsite laboratory, where we counted and recorded total fresh weight of each trees production. Seeds were subsequently dried at 105C, and after 24 hours, seed dry weight was recorded, and water content of seeds as a percent of fresh weight calculated. Measured seed production for the first three years likely underestimated total annual seed production due to noncontinuous sampling. To allow for graphical presentation of average see d production per tree, we calculat ed the percentage of each years seed production that occurred outside observed 5and 17week peaks during 200708 and 200809, respectively. We then added the average off peak seed production (converted to seed number and weight) to estimate total annual seed production measured for 20042007. These calculations assume that we captured the true peak of seedfall, which is corroborated by our year round phenological observations. An average of 1% and 18% of total seedfall in 200709 occurred outside

PAGE 31

31 of t he 17 week peak period in occasionally inundated and terra firme forest, respectively. An average of 24% and 40% of total seedfall in 200709 fell outside of the 5 week peak period in occasionally inundated and terra firme forest, respectively. We converted measured individual seed production to per hectare estimates by multiplying the average seed production per tree by the average number of trees with dbh > 15 cm per hectare for each forest type, the size when we found trees were likely reproductive (Klim as et al. 2007). Seed Removal and I nfestation Seed counts below trees did not include seeds that had been removed during the week long interval between collection visits. Seed removal before visits indicates a potential revenue loss, though not necessarily an ecological loss; removed seeds include eaten (predation) and dispersed seeds (VanderWall et al. 2005). Determining seed fate was outside the scope of this study, but it is ecologically relevant for species maintenance. We were interested in estimating the extent of seed loss during the course of a week, to determine whether more frequent seed collection trips were warranted. In addition to our weekly quantification of seed production, in 20072009 we used a second method to estimate seed production by c ollecting and counting all fruit husks below parent trees. Fruit husk counts have the potential to better represent actual seedfall in species with high seed removal: the fibrous capsules decompose slowly, and each mature seed leaves a clear mark within t he husk (Forget 1996). We counted the total number of fruit husks (including those for intact fruits) and d ivided by 4, the mean number of husks per fruit, to estimate fruit numbers. To estimate the number of seeds per fruit, we counted the number of seeds for all intact fruits or imprints for those that could be mounted in the lab. We obtained 315 such intact fruits in 200708 and

PAGE 32

32 341 in 200809. The mean number of seeds per fruit was multiplied by the husk based estimate of fruit number to estimate seed production prior to removal; this comparison was done for overall annual production without separation by forest type. This number was compared with measured seed production to estimate the extent of weekly seed removal. Hypsipyla ferrealis Hampson (Pyra lidae) moth larvae commonly attack C. guianensis seeds. Infested seeds can be recognized by 13 mm diameter holes created when larvae enter and exit the seeds (McHargue & Hartshorn 1983b). In some cases, larvae were found in relatively intact seeds, and in other cases, complete consumption of the endosperm was observed. Collected seeds were carefully examined for evidence of Hypsipyla ferrealis. Seeds with entrance holes were classified as infested, irrespective of hole size or stage of infestation. Data A nalysis Modeling variation in fruit production We modeled fruit production using a Poisson regression as a function of multiple predictor variables using the SAS procedure PROC GLIMMIX (version 9.2). Quantitative tree variables tested included dbh, tree height and crown cross sectional area. Dbh2 was also included based on research that found a quadratic relationship between dbh and fruit production for Bertholletia excelsa (Wadt et al. 2005). Class variables included (1) forest type, (2) vine cover and (3) abnormalities. Quantitative climate predictor variables also were tested, including total annual rainfall (from the year that flowering and fruit development occurred), rainfall during the annual 3month dry season from June to August prior to flowering, rainfall during the fivemonth wet season from January to May that preceded flowering, and the annual mean maximum temperature, mean

PAGE 33

33 minimum temperature and average temperature. Since seed production in each tree was measured weekly, a repeated measures st ructure was utilized, with the variancecovariance matrix structure specifically accounting for withintree correlation of observations. Pearson adjusted residuals were plotted to verify that the assumptions of Poisson regression were met. In all cases, h omoscedasticity and normality of the residuals were verified. Nonsignificant variables were dropped sequentially (based on p values), using a backward stepwise model selection procedure to determine our final model. We also ran the model using 5week peak seedfall periods each year to confirm that sampling methods (different temporal measurement periods between years) did not affect the model results. If results were significant in both models (all data and 5week peak seedfall model) and the direction of the effect was the same, we assumed that significance of results was not an artifact of sampling methodology. If this was not the case, or results were only marginally significant, we verified the direction and magnitude of interactions by looking at plots of LSM and performing multiple comparison tests using Scheffes method. We used the SAS procedure PROC CORR to identify and potentially eliminate model variables that were correlated. Tree height was correlated with dbh (pearson correlation coefficient (PCC) = 0.64), dbh2 (PCC = 0.60) and crown crosssectional area (PCC = 0.59), and was thus eliminated from the model because these multiple correlations could potentially lead to multicollinearity problems. Moreover, dbh is an easily and commonly measure d variable whereas height measurement is often problematic in dense tropical forest. Total yearly rainfall also was dropped from the model. It was correlated with both dry season rainfall (PCC = 0.66) and rainfall during the 5month rainy season (PCC = 0. 63), the latter two being more

PAGE 34

34 biologically relevant drivers in determining flowering and fruiting than the more general yearly mean. Finally, we dropped mean temperature which was correlated with maximum (PCC = 0.86) and minimum temperatures (PCC = 0.76). Synchronicity in fruiting We used Kelly and Sorks (2005) methodology to calculate three measures of seed production variability. First, to estimate variability in individual tree seed production (xCVi), we calculated the coefficient of variation (CV) of seeds tree1 for each tree over the 5year study period. We then took the mean of these individual tree CVs and calculated variability in seed production at the population level (CVp) by taking the mean number of seeds tree1 for all sample trees over the 5 year period. Finally, we calculated synchronicity by calculating the Pearson correlation coefficients for all possible pairs of trees in the sample and calculating the mean of these CVs (xPCC) (Snook et al. 2005). The entire sample population from both forest types was used to calculate synchrony. Results Phenology C. guianensis individuals displayed two flowering peaks over a 12month period in 20072009. One peak initiated during the height of the dry season and concluded at the start of the rainy s eason, followed by a second smaller peak during the rainy season (Figure 2 1).

PAGE 35

35 Figure 21 Percentage of individuals with new leaves (top panel) flowering (middle panel) and with observed fruitfall (bottom panel) in terra firme and occasionally inundated forests superimposed on monthly precipitation values (mm).

PAGE 36

36 Figure 21. Continued

PAGE 37

37 Figure 21. Continued

PAGE 38

38 As the first set of flowers matured into fruits over 4 months, some trees flowered again joined by trees that had not yet flowered that year; a lower percentage of trees flowered during the second peak (Figure 2 1). Trees in terra firme forest comprised the majority of individuals flowering and fruiting during this second peak, whereas very few trees in the occasionally inundated forest flowered d uring this later peak (Figure 2 1). The percentage of flowering individuals increased in concert with increase in precipitation that marked the beginning of the rainy season in all years (Figure 2 1). In 2006, the greatest percentage of flowering individuals in occasionally inundated forest occurred 1month after the peak in terra firme forest. Otherwise, we observed the greatest number of individuals with flowers in both forest types concurrently for all other years, though this period lasted longer in oc casionally inundated forest than in terra firme forest in 2007 (Figure 2 1). Fruitfall commenced 4 months after the onset of flowering, near the end of the rainy season, and mirrored the dual peaks seen in the flowering phenophase (Figure 2 1). Peak percentage of individuals with falling fruit was consistently timed close to the time of the annual peak in rainfall. An anomalous high rainfall period at the end of the first fruitfall peak, however, was observed in 2009 (Figure 2 1). We observed no difference in fruiting peaks between forest types. Seed P roduction Seed production, or the average number of seeds produced per tree, varied greatly by year with the highest years producing over 9 times more than the lowest years (Figures 2 1 & 2 2). While seeds wer e collected over more months in 20072008 and 20082009, even when off peak seed production estimates were added to the measured seed production for the first three years, the average number of seeds

PAGE 39

39 produced per tree was much greater in 200708 and 20080 9 (Figure 2 2). This was true in both occasionally inundated and terra firme forests. Figure 22 Average number of Carapa guianensis seeds per tree. Estimates include measured seed production data plus an estimate of nonmeasured seed production for th e first three study years, when seed production was quantified only during a 5 to 17week peak period. Forest type, various tree attributes, and several climatic cues explained seed production variation for our 5year study (Table 2 2 ). Seed production was significantly different between forest types (p < 0.0001) (Table 2 2 ) with higher seed production in occasionally inundated forest. Results differed based on the 5week peak seedfall model, indicating that forest type was not a significant predictor of s eed production (p <

PAGE 40

40 0.0967). When we further explored forest type effect in the 5week model, however we found significantly higher seedfall in occasionally inundated forests (Table 2 3 ), but only in years 200708 and 200809. Seed production in 20082009 in both forest types was higher than seed production in other years (either forest type). Seed production in occasionally inundated forest in 20072008 was also higher than seed production in the three prior years (both forest types) as well as sameyear t erra firme seed production (Figure 2 3). Trends indicated that trees in occasionally inundated forest produced more seeds per tree than those in terra firme forest (Figure 2 2). The proportion of trees with seeds was also higher in occasionally inundated f orest in 20072008 and 20082009 with over 80% of selected trees producing seeds. The percentage of trees producing in terra firme forests was approximately 50% in 20072008, equal to the 20042005 percentages, although mean seed production per tree in 2004 2005 was lower. Table 22 Results explaining variation in seed production, calculated using PROC GLIMMIX to perform a Poisson regression Variable Denominator DF Estimated parameters F value Pr > F Intercept 103 58.5281 22.48 0.0001 Forest type (OI) 1 03 0.4502 22.60 <0.0001 Dbh 103 0.3394 136.59 < 0.0001 Dbh 2 103 0.0041 110.35 < 0.0001 Crown cross sectional area 103 0.0040 6.91 0.00 99 Vine load (0) 103 0.6584 27.78 < 0.0001 Dry season rainfall (3 months) 4 0.03178 48.88 0.00 22 Wet season rainfa ll (5 months) 4 0.0209 92.67 0.0038 Mean maximum temperature 4 0.5932 36.36 0.0007

PAGE 41

41 Table 23 Results of the model explaining variation in seed production when only 5week peak seedfall periods were used for all years. (For brevity, estimated parameters for year and year x forest type interactions are not shown) Variable Denominator DF Estimated parameters F value Pr > F Intercept 104 9.2197 10.56 < 0.0 001 Forest type (OI) 1 04 1.4163 2.81 <0.0 9 67 Dbh 104 0.5330 121.16 < 0.0001 Dbh 2 104 0.00641 95. 25 < 0.0001 Vine load (0) 10 4 0.9647 23.96 < 0.0001 Year 4 107.75 < 0.0003 Year*Forest type 4 11.79 < 0.0176 Figure 23 Relationship between d.b.h. and seed production using least squared mean values.

PAGE 42

42 Table 2 4 The average number of seeds per hectare ( _x ) and standard error (SE) for Carapa guianensis in kilograms Year Terra firme Occasionally inundated Seeds per hectare ( _x SE) Seed weight per hectare (kg) ( _x SE ) Seeds p er hectare ( _x SE) Seed weight per hectare (kg) ( _x SE) 2004 2005 445 180 1.11 1.485 919 343 8.24 3.25 2005 2006 176 98 0.93 3.90 287 174 3.45 0.96 2006 2007 157 98 0.18 1.30 383 201 3.06 2.49 2007 2008 1995 727 24.04 7.98 6264 1044 80.26 10.73 2008 2009 4 5 75 935 55.96 0.09 104 21 1 776 121.83 0.19 Estimates of seed production per hectare varied broadly from hundreds to thousands of seeds, depending on the year (Table 2 4 ). We did not add off peak seed production to these numbers; our seed production per hectare underestimated seed production in the first three years. Seed production per hectare (whether measured by seed number or weight) was higher in occasional ly inundated forest than in terra firme forest due to greater seed production per tree, and a higher density of trees per hectare. Still, variability was high in both forest types. Multiple other measured variables also explained C. guianensis seed produc tion, inlcuding dbh, dbh2, crown crosssectional area, vine load, dry season rainfall, wet season rainfall and mean maximum temperature (Table 2 2 ). Overall, tree crown cross sectional area had a slightly positive impact on seed production (p = 0.0099). Al though we assumed that this relationship would have been stronger without the confounding effect of nonproducing trees, this was not the case; removing nonproductive trees resulted in a weaker relationship. Crown cross sectional area, however, was not si gnificant in the model that used only 5week peak seedfall (Table 2 3 ). The relationship between dbh and seed production was also significant and slightly positive (p < 0.0001). There was a quadratic relationship between dbh and seed production: seed production increased with dbh until 4050 cm,

PAGE 43

43 then decreased (Figure 2 3 ). Vine load was negatively correlated with seed production, such that those trees with > 75% crown vine load had significantly less seed production (p < 0.0001). None of the interaction t erms were significant in explaining seed production variation and were not included in the final model. Figure 24 Relationship between year and seed production using least squared mean values. Rainfall during the 5month wet period from January to May was negatively correlated with seed production. This was also true for mean maximum temperature. Conversely, rain measured during the regions 3month dry season was positively correlated with seed production.

PAGE 44

44 Seed Removal and I nfestation Based on the number of fruit husks and an average seed number per fruit of 8.9 3.7 (SD), we estimated a 20072008 preremoval seedfall of 26,010 10,754 seeds. The actual number of seeds collected and weighed that year was 28,373. In 20082009, the average seed number per fruit was 8.8 3.2. Thus, we estimated preremoval seedfall at 51,541 18,742 seeds. The actual number of seeds collected in 20082009 was 53,557, within the range of variation of our husk based estimates of seedfall. Seed infestation by H. ferr ealis in 20072008, was: 14.6% and 15.5% for occasionally inundated forest and terra firme respectively. In 20082009, infestation was 12.0% and 11.6% for occasionally inundated forest and terra firme respectively. Synchronicity in Fruiting Variability in seed production at the population level was moderate (CVp = 1.25). Individual variation was also moderate (xCVi = 1.24). The mean pairwise Pearson correlation coefficient, however, indicated low seed production synchrony in this population of C. guian ensis (xPCC = 0.276). Discussion Using the tropical tree Carapa guianensis as a model species, this study demonstrates that multiple variables interact at different scales to influence seed production. At the largest scale, climatic cues (rainfall and temperature parameters) were central to setting overall patterns in phenophases and seemed to best explain why high seed production years were consistent across both forest types (or habitats) examined. Habitat heterogeneity, including species abundance, spacing between individuals (Klimas et al. 2007) and underlying abiotic factors may be the primary reason for significantly higher seed production in occasionally inundated forest (F =

PAGE 45

45 22.60, Table 22), though climatic variables had a much stronger predic tor effect (Table 2 2) At the smallest scale, i ndividual tree attributes contributed to seed production heterogeneity within habitats; dbh, and cross sectional canopy area showed positive, quadratic relationships with seed production while vine load negat ively affected seed production irrespective of forest type. These central ecological findings can also help inform management of this economically important multiuse species. Annual Patterns in Flowering and Fruitfall We observed two periods of flowering and seed production during a 12month cycle in both forest types in 20072009. Initiation of flowering was closely tied to the climatic rainfall variable. Flowering occurred at the beginning of the rainy season during the first two years and peaked immediately prior to the initiation of the rainy season in the last two years of the study (Figure 2 1). Our findings are consistent with other research that indicates that flowering peaks in seasonally dry forests tend to come at the end of the dry or start of t he rainy season (Van Schaik et al. 1993) ; reproductive events generally occur during the period of low photosynthetic activity or after high rates of reserve accumulation (Fenner 1998) Indeed, Van Schaik et al. (1993) found that 67% of flowering peaks fel l within onemonth of the onset of the rainiest period for 53 tropical forest sites. This timing, prior to the onset of some of the seasons heaviest rains, may be conducive to pollination. Flowering near the onset of the rains is also strategic for reproductive success. Only 4 months separated the first evidence of flowers f rom the initiation of fruitfall (though see Fournier 2003); germinating seeds were thus well positioned for water dispersal and seedlings unlikely to be limited by available soil moistu re during germination, though excessive rains can increase seedling death due to fungal rot (RigamonteAzevedo personal communication).

PAGE 46

46 While there is evidence for flowering induction with a reduction in photoperiod that coincides with the rainy season (R ivera & Borchert 2001), we did not test this hypothesis: our dataset shows annual consistency in flowering times, the relative lack of synchrony indicates control of phenology by sitedependent temporal variation in water status as opposed to photoperiodic control (Borchert 1994). Episodic rains prior to the onset of heavier rains may have been the actual impetus for flowering; s everal species in Venezuela responded to episodic rains with increasing water potentials, flowering and new leaf production (Daz & Granadillo 2005). Individual trees, however, differed in flower initiation: some trees concurrently had buds, flowers, developing fruits and fruitfall. This va ri ation may promote genetic recombination if spatial nonaggregation of flowering trees forces longer distance pollen dispersal. The timing of flowering, however, is likely also related to the timing of vegetative phenology of this species: initiation of flowering coincided with the peak in the number of individuals with new leaves ( Figure 21 ). Rep roductive events require substantial carbon reserves and the increased photosynthetic capacity of plants after leaf flush may help provide the necessary reserves for this largeseeded species. Seed Production Variation by Forest T ype We found significant ly higher seed production per tree in occasionally inundated forest which was further increased by higher densities in this forest type, though this effect was not one of the strongest effects ( F=22.6, Table 2 2 ). C. guianensis has long been considered pre dominantly a species of flooded forests (Pennington 1981, McHargue & Hartshorn 1983a, Fournier 2003), though previous research found high densities of individuals in terra firme forest (Klimas et al. 2007). Unmeasured microenvironmental characteristics, s uch as soil moisture or nutrient availability, may favor

PAGE 47

47 survival and growth of this species in inundated environments and also explain some of the differences in seed production between these forest types. Density differences, however, might be the underl ying cause of seed production differences: increased spacing among trees in terra firme forest may lower pollination visitation, thus reducing seed set and subsequent seed production. While euglossine bees, an observed flower visitor of C. guianensis (Mau s 2006) commonly visit plants sequentially over a wide geographic area (traplining) (Janzen 1971), some large euglossine bees show local site fidelity (Ackerman et al. 1982). Influence of Forest Type Across Y ears C. guianensis seed production per tree var ied greatly between years (xCVi = 1.24, F=107.75, Table 23) Average seed production per tree in 20072008 was more than four times higher than average seed production during any of the first three years of our study, and average production in 20082009 w as more than nine times higher (Figure 2 2, Table 2 5 ). Large oscillations of seed output by individuals between highand low production years may be the rule among polycarpic plants (Herrera et al. 1998), though some studies have found that seed product ion on a populationlevel can remain relatively constant between years, even with high variation in individual production (Kainer et al. 2007). Our study showed that C. guianensis had variable seed production between years (CVp =1.25) and between individuals (xCVi =1.24). Indeed, the CVp for C. guianensis was higher than the median value of 1.10 for 108 tropi cal species (Wright et al. 2005), but below the CVp of a strict masting species (i.e. 1.6) (Kelly 1994), which often suspend populationlevel seed pr oduction in some years. Still, calculations of CVs for

PAGE 48

48 seed output are less reliable with less than 10 years of data (Kelly 1994) and these results could be improved with longer term monitoring. Table 2 5 Number of individuals sampled and average seed wei ght per tree ( _x ) and standard error (SE) for Carapa guianensis This table includes only measured seedfall; data shown do not include estimates of off peak seed production not measured in the first 3 years of this study. Terra firme Occasionally inundated Total Production n Seed wt tree 1 (kg) ( _x SE ) n Seed wt tree 1 (kg) ( _x SE) N Seed wt tree 1 (kg) ( _x SE) 2004 2005 20 0.12 0. 16 1 9 0.43 0.17 3 9 0.27 0 .11 2005 2006 24 0.10 0.42 30 0.18 0.05 5 4 0.14 0.19 2006 2007 23 0.02 0.14 29 0.16 0.13 5 2 0.10 0.10 2007 2008 51 2.59 0.86 5 3 4.19 0.71 10 4 3.41 0.56 2008 2009 50 6.03 0.01 53 6.67 0.02 103 6.36 0.01 The interaction be tween forest type and year was also significant in our peak seedfall model (Table 23). In this model, year was used to represent climatic variation rather than season rainfall or temperature. The latter variable types were measured annually, and there fore had to be excluded from this analysis to avoid over parameterization of the model and nonconvergence. Perhaps part of this interaction and likely the observed lower seed production in terra firme forest, can be explained by a 2005 fire in one of our terra firme plots that stressed trees and reduced stored reserves available for reproduction. This fire, however, coincided with an especially severe Amazonian drought that would have affected both forest types. Indeed, percentage of individuals flowering and total seed production were low in both forest types in 2005 and the subsequent year. Phillips et al. (2009) found a highly significant decrease in biomass immediately after the 2005 drought in 55 plots distributed throughout the Amazon basin (as compa red with the multi decadal period immediately

PAGE 49

49 prior); lower biomass production is linked to reduced reserves for reproduction and leaf flush (though see Huerte et al. 2006, Saleska et al. 2007). Samantha et al. (2010) found reductions in both PAR and surface shortwave radiation during the drought, further limiting species potential to rebuild stored reserves. Because we did not collect data prior to the drought, we are unable to determine whether our results represent natural seasonal fluctuation in phenology and seed production or whether they are partially the response of trees to severe drought stress. Variation by Tree A ttributes Individual tree attributes, particularly size, were also important in explaining seed production variation, an expected relati onship since vital demographic rates are normally stageor size dependent (Harper 1977). Indeed, the strength of the impact of dbh on seed production was higher than any of the other predictor variables (F=136.59). We found a quadratic relationship between seed production and dbh; trees in the middle diameter classes (30 < dbh < 50) were better producers than both smaller and larger trees (Figure 2 4), a relationship also evident for Bertholletia excelsa (Kainer et al. 2007). This contrasted with results f or the related Swietenia macrophylla and two other tropical trees, which showed a more linear increase in seed production with increasing diameter (Snook et al. 2005, NabeNielsen et al. 2009), though larger individuals had been eliminated from the study populations through logging, effectively eliminating the ability to detect production patterns of larger diameter trees. Our data revealed an expected, though relatively weak, negative association between vine load and seed production (F=27.78, Table 22) (Stevens 1987), a finding consistent for other tropical species (NabeNielsen et al. 2009, Wright et al. 2005, Kainer et al. 2006). Nonetheless, Londres (2009) found no significant effect of lianas on

PAGE 50

50 seed production for C. guianensis in a flooded forest near the mouth of the Amazon. Their sample, however, included very few trees with lianas and these were weighted heavily toward larger, better producing individuals. Influence of Climate C ues Mean annual maximum temperature was negatively correlated with C. guianensis seed production. Temperature may not have independently affected seed production in our study; the three years with the highest annual mean maximum temperature also had the least dry season rainfall (PCC = 0.43). Indeed, very few cases have demonstrated that phenological patterns are triggered by temperature and not by covariates, such as cloud cover, day length and solar elevation (Stevenson et al. 2008). Combined, these generate substantial seasonal variation in tropical irradiance (Van Schaik et al. 1993), often associated with temperature changes, which may affect tropical phenological patterns (Wright et al. 1999, Stevenson 2004, Zimmerman et al. 2007). Indeed, 2005 was characterized by lower available sunlight due, in part to significant aerosol loads in the atmosphere from biomass burning (Bevan et al. 2007, Koren et al. 2007). The negative association between temperature and seed prod uction, however, may have been due to plant drought stress: higher temperatures in 2005 coincided with t h e severe drought Recent research has shown an interaction between temperature and drought stress on plants (Adams et al. 2009, Allen et al. 2010) due, in part to increasing respiration rates with temperature (Hartley et al. 2006) but our study did not c ollect data to test this hypothesis. Threemonth dry season rainfall was positively correlated with seed production. Conversely, fivemonth wet season rainfall was negatively correlated with seed production. Slightly higher than normal dry season rains m ay allow trees access to a

PAGE 51

51 seasonally scarce resource, increasing stored reserves available for flowering and fruiting. Too much rain during the dry season, however, is likely detrimental. Threshold levels of drought synchronize flowering and enhance seed set in some tropical forest plants (Alvim 1960). Since a minimum flower abundance may be necessary to attract pollinators (Van Schaik et al. 1993), this drought influenced synchrony may facilitate pollinator activity. Our results correlating dry season rai nfall with seed production are not unique in the region; Kainer et al. (2007) found a highly significant, though weak, positive correlation between number of fruits of Bertholletia excelsa and 5month dry season rainfall. Rainfall data is also influenced by the occurrence of the El Nio southern oscillation. Wright et al. (1999) found that a 2year cycle of high, then low community level fruit production was observed twice for Barro Colorado Island, Panama, when a mild dry season followed an El Nio event. Zuidema & Boot (2002) found a significant drop in fruit production coinciding with low rainfall in an El Nio year. There may be a pattern with data from our study; the El Nio years 20042005 and 20062007 were associated with low fruitfall while the La Nia year of 20072008 was associated with high levels of seed production. Bertholletia excelsa trees in the region also had high levels of seed production in 2008 and 2009; and relatively low seed production in 2006 and 2007 (Wadt & Kainer, unpublished da ta), similar to the trends observed for C. guianensis though B. excelsa takes 14 months to mature from flower to fruit. The El Nio year in 2005 coincided with the severe drought caused by anomalous North Atlantic warming ( Marengo et al. 2008). If results from these species are an indication of

PAGE 52

52 basinwide trends, El Nio years or drought years may be associated with decreased seed production for tropical species Management C onsiderations While we observed high interannual variation in seed production, se ed production on a populationlevel occurred in all years, a positive implication for seed extraction and management. We found consistency in seed production in both forest types; high seed production in occasionally inundated forest coincided with high seed production in terra firme forest. In some years, harvest may not be viable in either forest type; years where expected harvests generate two to twelve kilograms of seeds per hectare (Table 2 4 ) may not generate sufficient revenue to justify time invest e d in seed collection. One way to reduce harvest costs is to focus on trees that show evidence of being topproducers (Kainer et al. 2 007, Snook et al. 2005); a subset of monitored C. guianensis trees in this study showed evidence of consistently superior s eed production. The percentage of plants flowering was a good indicator of future fruit abundance for C. guianensis a finding described in similar studies (Foster 1982, Heideman 1989). Anticipating years with high seed production prior to the onset of f lowering, however, may be difficult. This study does not have the temporal length necessary to predict patterns in seed production variability between years. Indeed, difficult to predict, variable cycles may be the norm for this species. Further monitoring however, may indicate a maximum number of high production years before the plant must rebuild resources. The two production peaks per annum observed in our study may facilitate management. Leaving seeds from the second smaller production peak in the for est may be sufficient to maintain species recruitment though this second peak was not

PAGE 53

53 forthcoming in low production years Results from population models often suggest that very low seedling establishment: rates as low as 1 3% for an Amazonian herb (Bruna 2003) are sufficient for positive population growth rates Londres (2004) found that seed production of C. guianensis far exceeds harvest levels for a community in the eastern portion of the Brazilian Amazon basin; the community harvested only 0.1% of the total estimated annual seed production in the area (Londres 2009). Excessive extraction in our study area is not likely in the short term. Current local market prices of R$1415 for a 60 kilogram mesh beansack of C. guianensis seeds and limited deman d in the state of Acre effectively limit interest in largescale collection. Though there was little evidence of seed predation by vertebrates in the study region, seed infestation by H. ferrealis larvae affected 11.6 to 15.5% of viable seeds. Leaving fal len seeds in the forest facilitates the rapid spread of larvae to noninfested seeds. Larval damage can be minimized by frequent seed harvest coupled with rapid storage and/or oil processing. Seeds can be stored prior to oil extraction, but the larvae must be killed either by drying seeds at high temperatures or submerging seeds in water for 24 hours (Ferraz 2003); solar drying spreads H. ferrealis infestation (RECA, personal communication). Still, infested seeds can be used for oil extraction. When we experimentally extracted seed oil via a small press, we included seeds with evidence of H. ferrealis entry with noninfested seeds. The quantity of oil extracted was dependent almost completely on the weight of the seed sample (r2=0.98) (Klimas, unpublished data), but seed quality was not tested. This study is an important first step in management of a species with variable inter annual fruiting, and can help predict years with high fruit production crucial for

PAGE 54

54 economic revenue. We recommend further study of t emporal seed production trends and variability as our study represents only a snapshot in the life of this long lived species.

PAGE 55

55 CHAPTER 3 VIABILITY OF COMBINED TIMBER AND NONTIMBER HARVESTS FOR ONE SPECIES: AN EXAMPLE USING CARAPA GUIANENSIS Introducti on There is growing consensus that sustainable use has a significant role to play in nature conservation globally ( Brechin et al. 2002, Heywood & Iriondo 2003, Nepstad et al. 2009). While preservation via protected areas is a cornerstone of conservation (B rooks et al. 2009) and an essential part of any strategy to conserve tropical forest diversity, protected areas inevitably include an incomplete set of the regional biota and are often embedded in a modified and fragmented landscape. Even if protected areas are locally effective, they encompass only 9.8% of tropical forests and 11.1% of Amazonia (Schmitt et al. 2008), often too small to be comprehensively effective ( Janzen 1986). Luckily, managed and secondary forests can also provide important conservation benefits, including provision of ecosystem services (Daily 1997, Bhagwat et al. 2008, Gardner et al. 2009) and species habitat for biodiversity conservation ( Lamb et al. 2005, Scales & Marsden 2008). Maintenance of these forested areas adjacent to or near by protected areas could complement protected areas by creating a forested mosaic at the landscapelevel. Recognizing and increasing the conservation value of lands outside protected areas is a necessary component of any long lasting conservation effor t (Harvey et al. 2008, Ros Tonen et al. 2008). In the Brazilian Amazon, existing legislation has the potential to promote forest cover on private land: Brazilian law currently requires maintaining 80% of a private landholding in the Amazon as forest reserv e (Medida Provisria 2001). Enforcement is often lax, however, suggesting that this legislation is not likely to maintain forest cover on privately owned lands while deforestation for

PAGE 56

56 conversion to other land uses remains a competing and economically attractive alternative. Managing forests for services and products that yield economic benefits is another way to enhance the value of standing forests while providing benefits to the landowner, thus encouraging forest persistence. This option has the benefit t hat economic incentives for management are applicable to diverse groups, including private landowners, forest residents, indigenous groups and smallholders charged with management of conservation units. Smallholders alone control over 16% of the Amazon (Ba rreto et al. 2005, Lima et al. 2006), more than the area currently under legal protection. Social and economic benefits derived from standing forests must be competitive with benefits derived from forest conversion to other land uses, however, since privat e owners are not expected to manage primarily in the public interest (Zarin et al. 2007). While market valuation of some benefits, like ecosystem services and carbon storage is still developing (Engel et al. 2008, Stickler et al. 2009), multiple forest re sources already provide concrete economic benefits, including timber and nontimber forest products (NTFPs). Logging provides clear economic benefits ( Verssimo et al. 1992, Lentini et al. 2003, AIMEX 2005), although it does not provide a continuous flow o f benefits unless it is specifically managed to do so (Oliveira et al. 1998, Oliveira 2000), which is rarely the case: indiscriminant oneentry logging remains the predominant form of timber extraction (Putz et al. 2000). Timber extraction can also provide conservation benefits: half of Borneos remaining forests (approximately 200,000 km2) that have active forestry concessions maintain significant wildlife conservation value (Meijaard et al. 2005) and biodiversity (Whitmore & Sayer 1992), making them a viable component

PAGE 57

57 of multi functional conservation landscapes (Meijaard et al. 2005). Logging, however is an intensive and extensive land use that has significant drawbacks: removing the stem causes significant forest disturbance (Verissimo et al. 1992), increases forest susceptibility to fire (Laurance et al. 1998, Cochrane et al. 1999, Alencar et al. 2004), and typically permits new forest access and conversion (Walker 2003, Fearnside 2005). In contrast, NTFP harvest can provide a more consistent (Pearce et a l. 2003) cash flow from harvesting a diversity of species over the course of a year and may or may not provide similar revenue levels as logging, but likely over longer time frames (i.e., decades). NTFP harvests are generally less destructive than logging, though harvest of stems, bark and apical meristems imply harvest of whole trees (Ticktin 2004). H arvest s are typically eas ier to sustain over years or decades and add to the perceived value of the forest ( Chopra 1993, Gunatilake et al. 1993, Marshall et al. 2003) but alone, may not provide enough revenue to promote forest maintenance. NTFP harvest can also affect persistence of populations ( Peres et. al. 2003, Peters 1996), alter forest structure ( Belcher & Schreckenberg 2007) and ecosystem processes at many levels (Ticktin 2004) and may be insufficient to lift communities out of poverty (Morsello 2006). Another option, combining these two types of forest products through multipleuse forest management (MFM) is increasingly envisioned as a preferable al ternative to timber dominant management models (GrciaFernndez et al. 2008). Limited research, however, has focused on the relative compatibility of coupling timber and NTFP harvests (though see Snook 2000, Guariguata et al. 2008, Menton et al. 2009, Gua riguata et al. 2010). Still thoughtful harvest of multiple forest products seems

PAGE 58

58 increasingly necessary to compete with relatively lucrative activities, like conventional logging and cattle ranching (Pearce 1996, Pearce et al. 2003). Determining ecolog ically sustainable harvest limits is necessary for developing appropriate forest management strategies, despite the significant challenge to produce effective simulations of tropical forests managed for both NTFPs and timber (Ribeiro do Valle et al. 2007). Matrix models are powerful tools useful for identifying population trends and responses to management regimes (Leftkovitch 1967, Crowder et. al. 1994, Olmsted & Alvarez Buylla 1995), setting harvest limits from a population that are compatible with its co ntinued existence (Nantel et al. 1996, Ratsirarson et al. 1996) and identifying key life stages as management targets (Crouse et al. 1987). In addition, matrix models can be used to simulate sustainable harvest levels for timber, NTFPs or a combination of management strategies for single species, an important factor in determining appropriate management strategies for the third to almost half of timber species that also have value for their NTFPs (Martini et al. 1994, Herrero Juregui et al. 2009). W e empl oy matrix models to examine sustainable use of Carapa guianensis one such multipleuse species. We question whether management of C. guianensis a tree species valued for both its high quality timber and its seed oil, can be managed sustainably for both s eed and timber harvests. Our objectives were: (1) to quantitatively assess the impact of seed and timber harvesting on C. guianensis in stochastically varying environments, (2) to determine whether the impact of harvest varied between occasionally inundated and terra firme forests, and (3) to simulate potential ecologically viable harvest scenarios of C. guianensis seeds, timber, or seeds and timber. We

PAGE 59

59 hypothesized that intensive seed collection would still result in sufficient seedling recruitment to ensure continued species persistence Wadt et al. (2007) found that high levels of seed harvest (up to 71%) were compatible with population persistence and simulation models indicate that extreme reductions in recruitment are necessary for population declines (Bruna 2003) We also hypothesized that terra firme forests would allow even more intensive seed harvests as seeds in this environment may be safesite limited. Klimas et al. (2007) found that C. guianensis seedling densities were almost three times higher in occasionally inundated forests than in terra firme forests perhaps due to the limited availability of suitable germination sites T herefore, collection of seeds that would otherwise perish would allow for increased seed harvest levels without impact ing future population viability. We examined sustainable harvests of C. guianensis in two of the forest types in which it occurs naturally, and at different densities and abundances (Klimas et al. 2007). Vital rates, key demographic variables that form the basis for matrix models, depend on individual size (Harper 1977), species distributions and abundances between habitats (Bruna & Kress 2002, Wagner & Fortin 2005), and habitat heterogeneity (Webb & Peart 2000, Svenning et al. 2006) due to variation in phy sical variables (Gartlan et al. 1986, Hubbell & Foster 1986, Davies et al. 1988, Lieberman et al. 1995, Whitmore 1984). Methods Study S pecies Mature C. guainensis trees can reach heights of 30 meters ( Ferraz et al. 2002 ) though neither tree height nor dbh reliably predicts age. Carbondating of two individuals in our study region estimated the age of a 37.5 dbh individual at 172 years while a 17 cm dbh individual was aged at 785 years (both of these ages are based on

PAGE 60

60 extrapolating growth rates determined by multiple radiocarbon measurements) (Vieira et al. 2005). Fruit fall coincides with the rainy season, with an onset around December and cessation by early June in our study region ( See Chapter 1). C. guianensis has no seed bank and germination occurs rapidly, often within two weeks of see d fall (Klimas, personal observation). Seeds are enclosed in a dehiscent capsule that normally breaks open with ground impact, freeing seeds for germination or dispersal via water or frugivores (McHargue & Hartshorn 1983, Plowden 2004). Carapa guianensis is an important timber tree in the Neotropics (McHargue & Hartshorn 1983, Mabberley 1987, Dayanandan et al. 1999). It is also valued for the high quality oil extracted from its seeds (Shanley & Medina 2005) which sells both locally and in international markets; Brazil exported between 200300 tons of oil annually between 1974 and 1985 (Clay et al. 2000). Pure C. guianensis seed oil is used for medicinal applications (Rodrigues 1989) with valueadded products including soaps, shampoos, candles and repellent torches (Shanley 2005) This species is considered to have such great economic potential that the Amazonian State of Acre, Brazil identified it as one of six priority species for extraction research (Acre 2000) Study S ite We carried out field measurements within four 400 m x 400 m (16ha) plots established within the 1200 ha experimental forest of the Brazilian Agricultural Research Corporation (Embrapa) in the eastern portion of the state of Acre ( latitude 10 01' 28" S and longitude 67 42' 19" W). Average annual temperature is 24.5C with a dry season from June to August (INMET 2009). The region has rolling topography including upland habitat ( terra firme ) and occasionally inundated areas. Two plots were established w here the majority of the environment was classified as terra firme forest and two in

PAGE 61

61 occasionally inundated forest (Klimas et al. 2007). All individuals > 10 cm dbh were inventoried in these plots and tree canopy position (dominant, codominant, intermediate or suppressed), spatial coordinates, and diameter at breast height (dbh) were measured. Individual Survival Rates and Diameter G rowth Growth rates used in the demographic model were based on the average growth rate of individuals in a given size class. We installed dendrometer bands on a stratified random sample of over 500 trees in 2003, controlling for location (forest type) and diameter at breast height (dbh) categories (Table 3 1). Following a 9month period of band adjustement annual circumference growth was monitored between May and June of 2004 through 2009, recording measurements during the dry season to reduce variance (Sheil 1995, 1997 but see Baker et al. 2003). A nnual circumference increment was divided by pi to convert to annual diameter increment T ree mortality of all marked individuals was recorded annually during growth measurements Table 31 Number of trees with dendrometer bands installed in 2003 and measured in 2004. N reflects original tree numbers sampled in 2004 prior to subsequent tree mortality. Forest type Diameter class (cm) N Occasionally inundated 10 to 20 104 20 to 30 70 30 to 40 39 40 to 50 20 50 + 15 Total 248 Terra firme 10 to 20 100 20 to 30 71 30 to 40 40 40 to 50 22 50 + 9 Total 242

PAGE 62

62 We monito red seedling and sapling survival and growth in 32 10 m x 10 m subplots within each of the larger 400 m x 400 m plots (Klimas et al. 2007). We noted x, y coordinates and annually monitored all seedlings (individuals < 1.5 m tall) and saplings (individuals > 1.5 m tall and < 10 cm dbh) in each subplot from 2005 to 2009, between May and July For seedlings, we annually measured basal diameter (soil level) and stem height; for saplings we measured dbh (Klimas et al. 2007) calculating annual growth increment s by increases in seedling height and sapling dbh, respectively. We measured smaller saplings using digital calipers, marking the measurement location with a permanent marker to ensure year to year consistency. For larger saplings, we sprayed paint around a circular piece of metal the same size as the diameter tape, taking annual measurements by placing the diameter tape within the nonpainted area. For seedlings and saplings, we calculated survival based on observed mortality each year in each forest type. For seedling growth, we pooled individuals from both forest types to calculate annual growth, because high mortality substantially decreased the approximately 900 seedlings tagged over the 5year study Likewise, survival of o nly 76 saplings did not leave representative sam ple sizes for calculating growth by year and forest type and thus sapling measurements were also pooled across forest types Parameter Estimation for Transition M atrices The pop ulation was classified into eight size classes considered biologically relevant : seedlings (individuals < 1.5 m tall) ; seedlings that regressed in growth but survived (also <1.5 m tall) ; saplings (individuals > 1.5 m tall and < 10 cm dbh) ; 10 cm dbh < 20 cm ; 20 cm ; 30 cm ; 40 cm ; and dbh Our justification for the given size classes was based on differences in size specific mortality and fertility. Mortality rates were very high for all newly

PAGE 63

63 germinated seedlings with survival increasing substantially when seedlings reached a height of approximately 1.5 m Individuals in the 10 20 cm dbh class had a low probability of producing seeds (Klimas et al. 2007); dbh was then quadratically relat ed to seed product ion (Chapter 2) Every year, we observed a significant fract ion of seedlings that regressed. I n most cases, these seedlings were dead the followi ng year, but some survive d Since seedling cannot regress back into the seed stage (Figure 31) w e divided our fertility entries of average sizespecific seed ling production into the first and second rows of the matrix based on the proportion of seedlings that grew or regressed, respectively (Figure 3 2) Figure 31 T he life cycle of C. guianensis with straight arrow s indicating the probability that an individual will grow (Gi) from one stage to the next during time interval t Curved arrows indicate the probability that an individual will survive and remain in a given size class (Pi) during t ime interval t Starred stages can produce seeds, though not all individuals or stages produce seeds in any given year. Numbers represent the midpoint of dbh size classes. Seeds are not included in the matrix as they germinate rapidly and become either see dlings that experience positive growth (Seedlings1) or seedlings that experience negative growth (Seedlings2) during the oneyear time interval t

PAGE 64

64 Seedling 1 Seedling 2 Sapling 3 15 4 25 5 35 6 45 7 55 8 Seedling 1 P 1 1 G 1 0 0 F 4 *P pos F 5 *P pos F 6 *P pos F 7 *P pos F 8 *P pos Seedling 2 0 P 2 2 G 2 0 F 4 *P neg F 5 *P neg F 6 *P neg F 7 *P neg F 8 *P neg Sapling 3 G 1 = 1 *(g 1 /c 1 ) G 2 = 2 *(g 2 /c 2 ) P 3 3 G 3 0 0 0 0 0 15 4 0 0 G 3 = 3 *(g 3 /c 3 ) P 4 4 G 4 0 0 0 0 25 5 0 0 0 G 4 = 4 *(g 4 /c 4 ) P 5 5 G 5 0 0 0 35 6 0 0 0 0 G 5 = 5 *(g 5 /c 5 ) P 6 6 G 6 0 0 45 7 0 0 0 0 0 G 6 = 6 *(g 6 /c 6 ) P 7 7 G 7 0 55 8 0 0 0 0 0 0 G 7 = 7 *(g 7 /c 7 ) P 8 8 G 8 Figure 32 Matrix entry calculations, where: Fi= average sizeclass specific fertility ; i =size class specific survival probability ; gi = size class specific growth rate (mm/year) ; ci = category width (100mm for most size classes, excluding seedlings and saplings which differed in their category width limits) ; Pneg = probability that a seedling will experience negative growth its first year ; Ppos = probability that a seedling will experience positive growth its first year ; Seedling1 = seedlings that experience positive growth during their first year after germination ; Seedling2 = seedlings that experience negative growth during their first year after germination

PAGE 65

65 We used the SAS procedure PROC LOGISTIC ( version 9.2) to calculate logistic regressions of survival and growth against size for individuals 10 cm db h ; we calculated survival and growth values taken from these fitt ed regressions at the midpoint of each size class, using logistic regression implemented in the R programming language (R Development Core Team 2006) to calculate the probability of growing from stage i to stage i +1 in the transition matrix. Combining the data from all size classes to estimate logistic regressions of survival and growth minimizes errors from small sample size in larger size classes; estimates of survival and growth for classes with few marked individuals are susceptible to chance variation around the true value (Caswell 2001, Morris & Doak 2002). Growth and survival curves were calculated: 1) for the entire dataset combining data from all years and both forest types, 2) per year using pooled matrices of individuals from both forest types and 3) annually for each forest type. We calculated transition probabilities by multiplying the probability of survival (Pi) by the probability of growth (Gi) from i to i +1 (Figure 32) We calculated the probability that a given individual would transition from stage i to i +1 as the inverse of the average duration in a given stage ( Caswell 2001), based on average sizespecific growth from dendrometer band measurements. Seedling and sapling transitions were the inverse of the average duration in a given stage based on height growth and diameter growth, respectively (Figure 3 2) Fertility values were based on the average seed production of an individual in a given size class converted to observed seedling frequencies. We quantified weekly size class specifi c seed production by collecting fallen seeds under a random subset of

PAGE 66

66 trees in both forest types. We used a combination of measured annual seed production from 07 08 and 08 09 and estimated seed production (for methodological details, see Chapter 1) Ma tr ix M odeling We compared transition matrices between years and forest types, using population growth rate and bootstrapped estimates of variability to determine whether matrices were the same (indicating that we should pool all matrices before running matri x projections) or whether differences between matrices, either between years or forest types, warranted using separate annual transition matrices for each forest type. Confidence intervals for each estimate of were calculated using bootstrapping: we reca lculated average growth in each size class each year based on a random reselection of individual growth rates. The upper and lower 95% confidence intervals of 2000 bootstrap runs were estimated using bootstrap percentile intervals (Caswell 2001) (Table 33 ). We used MATLAB ( 2002) for all analyses unless otherwise stated. Population m atrix models use the equation n(t+1) = An(t), where n(t) and n(t+1) are column vectors that contain the populat ion structure at time t and t+1, and A is a square m x m matrix containing the transition probabilities among categories during one year (Caswell 2001). In our case, m = 8 for all matrices in both forest types, where m is the number of size classes. In this equation, t he population structure at time t+1 is obtained by multiplying the structure at time t with the transition matrix that contains information on the dynamics of the population. When repeating this process many times, the population structure (relative to the initial population structure) and population gro wth rate become stable. In this situation, the rate at which the population grows (the asymptotic population growth rate) is equal to the dominant eigenvalue ( ) of matrix A,

PAGE 67

67 an inherent characteristic of the transition matrix. The stable population structure is equal to the right eigenvector of the matrix A (Caswell 2001). The long term behavior of the population is determined by If 1, the population is viable and is stable or increases; if < 1, the population is not viable and declines. Elasticity Calculations and R anking We also calculate d elasticity values for the vital rates underlying each 8 x 8 transition matrix. Elasticity analysis is often used to demonstrate the sensitivity of population growth rate to variations in vital rates (survival, growth and reproduction), allowing for analysis of effectiveness of management interventions at changing population growth ( Rogers Bennett & Leaf 2006, Carslake et al. 2009). While most studies have calculated deterministic sensitivities (Horvitz & Schemske 1995, Pascarella & Horvitz 1998, Meekins & McCarthy 2002), we calculated vital rate sensitivities since each matrix element is compose d of underlying vital rates, each of which may contribute to multiple matrix elements (Franco & Silvertown 2004, Morris & Doak 2005) E proportional, infinitesimal change in a vit al rate holding all other elements constant; elasticity (eij) is the slope of log plotted against log aij (Caswell 2001). The elasticities of with respect to a vital rate are often interpreted as the contributions of each of the vital rates to sin ce the elasticities of with respect to the vital rate always sum to one ( De Kroon et. al. 1986 ). We used vital rate means and a matrix definition to calculate deterministic elasticities of to vital rates, using symbolic functions to take derivatives (based on a modification of code by Morris & Doak 2002). We also ranked elasticities for each years model in each forest type.

PAGE 68

68 Modeling Environmental S tochasticity We incorporated environment al stochasticity into simulated population projections using i ndependent and identically distributed (iid) environmental states using MATLAB code based on Morris & Doak (2002) We assumed that the 5 years of our study (4 transition matrices) represented the range of typical conditions, each of which was equally likely to occur in the future; for each year 1 of the 4 matrices was chosen at random for each forest type and used to simulate population growth. To calculate the mean stochastic population growth r s) and associated 95% confidence intervals, we simulated population growth over 50,000 years. We used Tuljapurkars s of C. guianensis based on the assumption that the variation among the annual matrices was not large. Sustainable Harvest S imulations We used computer simulations to explore possible sustainable harvesting regimes by varying the percentage of seeds collected annually and the number of C. guianensis individuals harvested for timber. We simulated seed harvest by modifying the transition matrix such that between time t and time t + 1 fecundity values were proportionally reduced, with harvest levels varying from 50% to 95% (i.e. the fraction of seeds remaining after the simulated harvest varied from 50% to 5%). For timber, w e harvested only individuals dbh the legal minimum diameter limit for C. guianensis in the Brazilian Amazon ( Resoluo N. 406) and our largest size class, using a 25year harvest cycle. Although this cutting cycle is likely unsustainable (Gardingen et al. 2006), particularly given that C. guianensis is a slow growing species ( Vieira et al. 2005 ) we used the 25year cutting cycle to reflect on the -

PAGE 69

69 ground harvest realities Since timber i s normally logged during the dry season from June to August (INMET 2009), we simulated post reproductive timber harvests (after all seeds had fallen, and a percentage were harvested) ( See Chapter 1). We also ran simulations with annua l seed harvests combin ed with 25year timber harvests. All simulations were run under the stochastic environmental conditions simulated using i ndependent and identically distributed (iid) environmental states Quasi Extinction Threshold C alculations We also determined the exti nction time cumulative distribution function (CDF) the single most informative metric of a populations viability (Morris & Doak 2002). Since s is a long term metric of population growth or decline under stochastically changing environmental conditions, the CDF provides us with an estimate of the probability that a population will fall below a certain number of individuals in the short term. Fo r our simulations, we used a time horizon of 500 years and a quasi extinction threshold of 100 indiv iduals, equivalent to approximately three individual s dbh 10 cm per hectare. These simulated densities would be a drastic reduction f rom our original dens ities of 25.7 and 14.6 individuals dbh 10 cm per hectare in occasionally inundated and terra firme forests, respectively (Klimas et al. 2007) The CDF was calculated based on 20 runs with 5,000 separate realizations of population growth (Morris & Doak 2002). Results Vital Rates and Parameter E stimation Pooled average diameter growth was 2.89 mm ( 2.76 mm SD) per year, suggesting that an individual would take an average of 35 years to grow to the next size class. Class transition periods varied broadly fr om 17 to 769 years, the range limits for individuals at the highest and lowest growth extremes, respectively. Size class

PAGE 70

70 dependent growth followed similar patterns; annual diameter increment was low with high variability ( Figure 3 3 ). There was also variat ion in growth between years with a comparatively higher average growth in the first transition years (0506), which was consistent throughout all dbh classes and in both forest types. Growth trends were not significant, however, due to high variability. The sample size used to calculate annual growth increment decreased slightly over the study period due to tree mortality and dendrometer band damage Matrix entries indicating stasis (survival and maintenance in the same size class) were consistently high ( greater than 90%) for individuals dbh 10 cm in both forest types. Overall seedling survival was consistently higher across all years in terra firme forest, while sapling survival was lower across all years (Table 3 2). Higher fecundity values in 0708 and 08 09 were due to the variable annual seed production of C. guianensis ; fecundity values were also significantly higher in occasionally inundated than in terra firme forest (See Chapter 1). Average fecundity per size class was highly variable between years. This was, in part, due to temporally variable seed production, but individuals suspending seed production also dropped the contribution of certain size classes in some years. This was most obvious in the largest size class; individuals in this size cl ass (dbh in only two of the four measured years (08 09). Matrix Modeling We constructed population projection matrices for each forest type in 0506, 0607, 0708 and 08 09. Upper and lower confidence intervals (CIs) represent bootstrapped estimates of variability from random selection of individual growth rates. We found differences between transition matrices between years and forest types.

PAGE 71

71 annual matrices; in terra firme 3). Confidence intervals around were tight (Table 33), therefore we used four separate annual transition matrices for each respective forest type in all our simulation m odels. The differences between the forest types were great enough and the uncertainty around the estimates small enough to warrant different transition matrices ( Table 3 2 ). A Figure 3 3 Annual sizeclass dependent diameter growth (mean + SD) of C. g uianensis in terra firme forest (a) and occasionally inundated forest (b) Years included are 20052006, 20062007, 20072008 and 20082009.

PAGE 72

72 B Figure 33. Continued Elasticity Calculations and R anking The overall importance of types of vital rates was fairly consistent in both forest types (Figure 34). Stasis, the relative probability of surviving but remaining in the same size class, had the highest proportional contribution to population growth rate. Fecundity values had some of the lowest proporti onal were also low. Results from the pooled matrices in each forest type show general elasticity trends. Stasis had the greatest effect on population growth rate and stasis of different size classes consistently dominated the top four elasticity ranks, though the relative ranks of stasis of different size classes varied between years and forest types (Table 34) Stasis not only dominated elasticity rankings, but also stasis of the four top

PAGE 73

73 Table 3 2 Leftkovitch or size dependent, demographic matrices for Carapa guianensis in two forest types: occasionally inundated and terra firme For each forest type, there are four annual transition matrices. Matrix entries are: seedlings, saplings, 10 cm dbh Size class Year 1: Occasionally inundated forest Seedling 1 0.2166 0 0 0 0 49.6665 83.0504 0 Seedling 2 0 0 0 0 0 43.5165 72.7664 0 Sapling 0.0113 0 0.9638 0 0 0 0 0 10 20cm 0 0 0.0027 0.9643 0 0 0 0 2030cm 0 0 0 0.0326 0.9515 0 0 0 3040cm 0 0 0 0 0.0421 0.9404 0 0 40 50cm 0 0 0 0 0 0.0464 0.9253 0 > 50cm 0 0 0 0 0 0 0.0475 0.9438 Year 2: Occasionally inundated forest Se edling 1 0.2359 0 0 0 1.5885 8.5099 158.2206 0 Seedling 2 0 0.1581 0 0 1.1767 6.3036 117.2005 0 Sapling 0.0126 0.0085 0.9052 0 0 0 0 0 1020cm 0 0 0.001 0.965 0 0 0 0 20 30cm 0 0 0 0.0229 0.9588 0 0 0 3040cm 0 0 0 0 0.0318 0.9588 0 0 40 50cm 0 0 0 0 0 0.0318 0.9588 0 > 50cm 0 0 0 0 0 0 0.0355 0.9956 Year 3: Occasionally inundated forest Seedling 1 0.2843 0 0 1.4786 17.3035 12.5663 14.9277 4.8205 Seedling 2 0 0.0457 0 4.2509 49.7475 36.1281 42.9173 13.859 Sapling 0.0116 0.0018 0.9633 0 0 0 0 0 10 20cm 0 0 0.0021 0.9711 0 0 0 0 20 30cm 0 0 0 0.018 0.9597 0 0 0 30 40cm 0 0 0 0 0.0271 0.9495 0 0 4050cm 0 0 0 0 0 0.0344 0.9416 0 > 50cm 0 0 0 0 0 0 0.0387 0.9761

PAGE 74

74 Table 32. Continued Year 4: Occasionally inundated forest Seedling 1 0.1845 0 0 19.0448 55.2516 76.3062 100.9983 45.8275 Seedling 2 0 0.2204 0 3.1441 9.2085 12.7176 16.833 7.6379 Sapling 0.0163 0.0195 0.962 0 0 0 0 0 10 20cm 0 0 0.0022 0.9672 0 0 0 0 2030cm 0 0 0 0.018 0.9617 0 0 0 30 40cm 0 0 0 0 0.0218 0.9552 0 0 40 50cm 0 0 0 0 0 0.0263 0.9476 0 > 50cm 0 0 0 0 0 0 0.0317 0.9769 Size class Year 1: Terra firme forest Seedling 1 0.2444 0 0 0 4.7672 4.8234 40.682 0 Seedling 2 0 0 0 0 2.7241 2.7562 23.2468 0 Sapling 0.0155 0 0.8072 0 0 0 0 0 10 20cm 0 0 0.0052 0.9216 0 0 0 0 20 30cm 0 0 0 0.0334 0.9083 0 0 0 3040cm 0 0 0 0 0.0396 0.8989 0 0 40 50cm 0 0 0 0 0 0.0407 0.8893 0 > 50cm 0 0 0 0 0 0 0.0405 0.9186 Year 2: Terra firme forest Seedling 1 0.4848 0 0 0 4.8259 2.7075 18.7164 0 Seedling 2 0 0.1535 0 0 0.6894 0.3867 2.6737 0 Sapling 0.0414 0.0131 0.8404 0 0 0 0 0 1020cm 0 0 0.0033 0.9289 0 0 0 0 20 30cm 0 0 0 0.0185 0.9223 0 0 0 3040cm 0 0 0 0 0.0261 0.9185 0 0 40 50cm 0 0 0 0 0 0.0309 0.9171 0 > 50cm 0 0 0 0 0 0 0.0332 0.9513 Year 3: Terra firme forest Se edling 1 0.4878 0 0 0 0.7326 2.4643 4.348 1.9549 Seedling 2 0 0.0464 0 0 0.7326 2.4643 4.348 1.9545 Sapling 0.0121 0.0012 0.96 0 0 0 0 0 10 20cm 0 0 0.0029 0.9568 0 0 0 0 2030cm 0 0 0 0 0.0158 0.9782 0 0 30 40cm 0 0 0 0.0115 0.9773 0 0 0 4050cm 0 0 0 0 0 0.0203 0.9749 0 > 50cm 0 0 0 0 0 0 0.0247 0.9999

PAGE 75

75 Table 32. Continued Year 4: Terra firme forest Seedling 1 0.5312 0 0 0 6.477 9.4558 22.6187 24.9721 Seedling 2 0 0.2295 0 0 0.7623 1.1128 2.662 2.939 Sapling 0.0242 0.0105 0.9234 0 0 0 0 0 10 20 cm 0 0 0.0051 0.9337 0 0 0 0 2030cm 0 0 0 0.0205 0.9326 0 0 0 30 40cm 0 0 0 0 0.0247 0.9325 0 0 40 50cm 0 0 0 0 0 0.0277 0.9333 0 > 50cm 0 0 0 0 0 0 0.0297 0.9656 Table 3 3 Population asymptotic growth rates ( ) calculated from the Leftkovitch model for populations of Carapa guianensis in Acre, Brazil. Forest type Year Lower CI Upper CI Occasionally inundated Pooled 1.0029 1 1.0038 1.0037 1.0157 2 0.9956 0.9956 1.0070 3 0.9961 0.9912 1.0029 4 1.0167 1.0082 1.0399 Terra firme Pooled 0. 9706 1 0.9460 0.9380 0.9465 2 0.9550 0.9546 0.9595 3 1.0006 1.0005 1.0020 4 0.9841 0.9750 0.9850

PAGE 76

76 ranked size classes summed to proportional contributions of 0.84330.9222 and 0.7 205 0.9 692 in occasionally inundated and terra firme forests, res pectively. Figure 3 4 Elasticity values for Carapa guianensis vital rates in two forest types: occasionally inundated (A, C, E) and terra firme (B, D, F) For each forest type and vital rate, y axis values vary and there are four elasticity bars, one for each transition matrix. The first two panels (A, B) display the elasticity of to changes in the vital rate survival (s1s8). The second two panels (C, D) display the elasticity of to changes in the vital rate growth (g1g7). The last two panels (E F) display the elasticity of to changes in fertility, or the number of average seedlings produced by an individual in a given size class. The first five fertilities (f4 f8) represent seedlings that grow the first year; the second five fertilities (2f42f8) represent seedlings that shrink or experience negative growth the first year.

PAGE 77

77 Figure 34. Continued

PAGE 78

78 Figure 34. Continued

PAGE 79

79 Figure 34. Continued Table 3 4 Ranking of top four vital rate elasticity values in order of importance and their noted in parentheses. Occasionally inundated Terra firme Elasticity ranks Elasticity values Elasticity ranks Elasticity values Year 1 (1 5) 0.2 721 (15) 0.3816 (saplings ) 0.2693 (25) 0.2459 (25 ) 0.2063 (35) 0.1683 (35 ) 0.1648 (45) 0.1137 Year 2 (5 5) 0.5261 (1 5) 0.3303 (1 5) 0.1187 (2 5) 0.2611 (25) 0.1018 (35) 0.1710 (3 5) 0.0993 (45) 0.1351 Year 3 (15 ) 0.3443 (5 5) 0. 8931 (saplings ) 0.2643 (3 5) 0.0269 (25 ) 0.2139 (25) 0.0265 (35 ) 0.0807 (4 5) 0.0227 Year 4 (15 ) 0.3401 (5 5) 0.1917 (saplings ) 0. 3137 (1 5) 0.1888 (25 ) 0.1917 (25 ) 0.1829 (35 ) 0.0767 (saplings ) 0.1571

PAGE 80

80 Table 3 5 Simulated stochastic population growth rate ( s) with simulated 95% confidence interval s and stochastic growth as calculated by Tuljapurkar's T) for populations of Carapa guianensis in Acre, Brazil. Simulations were conducted with no harvesting, varying levels of seed harvest, timber harvest and combined seed and timber harvests. All transition matrices had an equal probablity of selection. Forest type Harvest s 95% Conf. Int. T Occasionally inundated No harvest 1.00 28 1.00 04 1.005 3 1.00 28 1 0% seeds 1.001 7 0.99 93 1.00 40 1.001 7 20 % seeds 1.0005 0.9982 1.0028 1.000 4 30% seeds 0.999 1 0.99 69 1.00 14 0.999 1 50% timber 25 yr cutting cycle 1.00 26 1.00 03 1.00 48 100% timber 25 y r cutting cycle 1.0022 1.000 1 1.00 44 10% seeds & 50% timber 1.0014 0.9992 1.0036 10% seeds and 1 00% timber 1.0012 0.9991 1.0033 Terra firme No harvest 0.9704 0.9678 0.9731 0.9704 10% seeds 0.9697 0.9670 0.9724 0.9696 20% seeds 0.9688 0.9661 0.9716 0.9687 30% seeds 0.9679 0.9651 0.9707 0.9678 50% timber 25 yr cutting cycle 0.9681 0.9659 0.97 02 100% timber 25 yr cutting cycle 0.9672 0.9651 0.9693 10% seeds & 50% timber 0.9671 0.9650 0.9693 10% seeds and 100% timber 0.9663 0.9642 0.9684 Modeling Environmental S tochasticity The simulated stochastic population growth rate s) in occasionally inundated s = 1.0028) and a declining population in terra firme s = 0.9704) (Table 3 5 ). The 95% confidence intervals for stochastic population growth indicated an increasing population in occas ionally inundated forest in the absence of harvest, but when even low levels of seed harvest (10%) were simulated, the lower CI dropped below one I n terra firme forest in contrast, both s and

PAGE 81

81 the upper bound of the simulat ed confidence interval remained below one even in simulations without harvest (Table 3 5 ). Tuljapurkars approximations corroborated simulated stochastic population growth rates (Table 3 5 ). Sustainable Harvest S imulations No levels of timber or seed harvest were considered viable in terra firme forest s < 1 though a major caveat of our study is that we have only 4 transition matrices for a species that can live hundreds of years (Vieira et al. 2005) If the observed environmental conditions were characteristic of expected future conditions, we would expect a decline in the terra firme population. Indeed quasi extinction simulations indicated that there was a 100% probability that the population would have less than 3 individuals dbh 10 cm per hectare after 93 years in the absence of either seed or timber harvest Multiple harvesting options, however, were available to sustainably manage C. guianensis in occasionally inundated forest. Simulations predicted population growth in occasionally inundated forest with harvest regimes of 10 % and 20% o f total seed production, but the lower 95% CI was below one for both of these harvest scenarios (Table 3 5 ). Timber harvests of both 50% and 100% of individuals with dbh 0 cm were also sustainable and the probability that the population would have fewer than 3 individuals dbh 10 cm remained zero for the 500 years modeled While we found this surprising since stasis had some of the highest elasticity values ; however modeled harvest levels were extremely low, involving fewer than 2 trees per hectare every 25years in occasionally inundated forests S ince smaller individuals (< 50 cm dbh) comprised a larg e er part of the per hectare densities, simulations did not reduce the population to fewer than 3 individual s dbh 10 cm per hectare until almost 1000 years in the future S imulations that limited recruitment while

PAGE 82

82 simultaneously harvesting larger individuals affected both recruitment of individuals into the smaller dbh classes and seed production in the larger dbh classes, leading to cumulative distribution functions ( CDFs ) reaching 1.0 after 150 years Low s eed harvest levels (10%) were also potentially compatible with both 50% and 100% harvest in the largest size class (dbh though the CDF indicated a 100% probability that these management scenarios would lead to fewer than 3 individuals dbh 10 cm per hectare during the 500 year time interval modeled. This model also does not include increased mortality accompanying harvest, and may overestimate the lower confidence bound (Table 3 5 ). Discussion Sustainable Stochastic Harvest Estimates in Contrasting Forest T ypes Our results indicate that some populations of C. guianensis may be suitable for extraction of seeds, timber, or a combination of both. High simulated harvests were not sustainable across all habitats, however. In our terra firme sites, demographic parameters suggested that no tested levels of seed and timber extraction were sustainable; quasi extinction simulations indicated pop ulation decline regardless of harvest levels While we hypothesized that safesite limitation would make seed harvests more viable in this forest type, the low survival of individuals dbh 10 cm made both seed and timber harvests non viable. While C. guia nensis does occur in terra firme forests, it is considered more competitive in forests with some flooding regime ( Pennington 1981, McHargue & Hartshorn 1983a Fournier 2003). Furthermore, our study region is close to the southwestern limit of the C. guiane nsis range, perhaps further compromising population resilience (Kawecki 2008)

PAGE 83

83 In contrast our modeling results revealed that a diversity of harvest regimes would be ecologically sustainable in occasionally inundated forests. In exploring sustainable see d harvests, removing 10% of the total seed production proved feasible and consistent with a stable or increasing population (Table 3 5 ) though the quasi extinction threshold of less than 3 individuals dbh 10 cm per hectare was reached within the 500yea r time interval High levels of seed harvest normally do not prevent species persistence (Bruna 2003) which led us to hypothesize that even higher levels of harvest would be viable. Indeed, 10% of seed harvest is relatively low in comparison with studies reporting sustainable seed harvest levels in excess of 50% of popul ation level production for Bertholletia excelsa (Zuidema & Boot 2000) and Sclerocarya birrea ( Emanuel et al. 2005) Nonetheless, c urrent market demand for C. guianensis seed oil is minimal in our study region, and would not likely support higher (> 10%) seed harvest levels. I n the mouth of the Amazon basin however, where C. guianensis is a major component of the dominant flooded forests and the seed oil market is moredeveloped demand is indeed high and Londres (2009) reported that still local seed production far exceeded the estimated 0.1% harvest levels We also found the potential for compatibility between seed and timber harvest in occasionally inundated forests. Our results suggested that 1 0% annual seed harvests were compatible with 100% timber harvests (of trees equivalent of approximately two trees per hectare. Our simulations were based on growth data from best fit logistic regressions but we expect that incorporating growth et al. 1998).

PAGE 84

84 M atrix models with limited classes can inflate values if a small proportion of individuals pass rapidly through the life cycle and reach reproductive size at a very young age, which has led some authors to recommend integral projection models (IPMs) to correct these unrealistically fast transitions (Zuidema et al. 2010, though see Ramula & Lehtil 2005 ). Nonetheless, we used regressions that related growth and size over the entire size range to calculate our average growth for each dbh class, a method that resembles the IPM approach (Zuidema et al. 2010). Finally, category width does not always inflate ; individuals that pass rapidly through the matrix but produce few recruits likely have less of an inflationary impact (Zuidema et al. 2010) though we are unsure if this is the case for C. guianensis Elasticity Calculations and R anking Elasticity anal yses indicate that stasis of varying size classes, measured by the vital rate survival, contributes the most to population growth rate. T his is consistent with (Pinero et al. 1984, Horvitz & Schemske 1995, Olmsted & Alvarez Buylla 1995, Morris & Doak 2005) and likely indicate that selection for high mean survival is higher than selection acting on any other vital rate (Morris & Doak 2005, Forbes et al. 2010) Even though high observed survival ( over 90% ) for individuals dbh (Table 3 2) could not be easily increased. Indeed, the lower stasis values measured for individuals in terra firme forests were partly responsible for the decreasing growth rate ( < 1) seen in the transition matrices for this forest type. Some of the increased mortality that led to reduced stasis values may have resulted from 2005 drought related fire (Marengo et a l. 2008) in one of the terra firme plots. Still, we also observed higher mortality of pole-

PAGE 85

85 sized trees in terra firme areas unaffected by fire (unpublished data), perhaps due to drier soil conditions associated with forest type. To explore drought effects, we changed drought year frequency from the observed four to every 10 years, whereby s increased to 0.9759 ( CIs 0.9731, 0.9787) If the observed drought represent s a rare catastrophe for this population, our iid model may severly underestimate s due to over inclusion of deleterious populationlevel effects especially for the drier ter ra firme forests. s was consistent in both forest types and throughout our 5year study period (Figure 3 4 ). C. guianensis is a long lived species (Vieira et al. 2005) however, and continued demographic monitoring may indicate higher importance of other vital rates during some years though see (Forbes et al. 2010) Indeed, Doak et al. (2005) found that the mean standard error of model predictions declined with higher sampli ng intensity and more importantly, duration. The high elasticity of survival in the largest dbh class in some years (Figure 3 4) also serves as a cautionary note for timber harvest. While we observed long term growth in the occasionally inundated forest ( s > 1) even with low levels of timber extraction, the periodic high elasticity of this dbh class merits prudence. Still, f ecundity and growth should continue to be considered in C. guianensis management strategies. Pfister (1998) found that vital rates w ith low elasticities tend to have high variance. Gaillard et al. (1998) notes that vital rates with high variation can have larger effects on population growth rate than rates with high elasticity. Favorable environmental conditions, however, may enhance m ultiple vital rates (Morris & Doak 2002). Years of high fruit production coincided with years of high growth for the related Swietenia macrophylla (Snook et al. 2005), though other studies have indicated a trade-

PAGE 86

86 off between seed production and growth in any given year for other species (Norton & Ke lly 1988, Koenig & Knops 2000). Limitations of the Harvesting M odels While five years of survival, growth and reproductive measurements would be adequate for some species, this study only begins to delve into the variation in environmental conditions likely experienced over the lengthy lifetime of a C. guianensis individual. We also found high variation in some of the vital rates: both seed production and growth were characterized by high variability around mean v alues. Large sampling variability in the estimated transition rates increases the uncertainty in the estimated matrix and quantities derived from it, such as the population multiplication rate and sensitivities of matrix elements ( Gross et al. 2006), thoug in sensitive to variations in these vital rates, in comparison with variations in survival. Our model did not include any spatial components though previous research has shown aggregation of C. guianensis especially juveniles (Klimas e t al. 2007). We also attributed the mean growth within a dbh class to all individuals instead of modeling individual growth. There is some evidence that individual trees with a history of rapid growth contribute disproportionately to population growth (Zui dema et al. 2009), but this phenomenon does not easily translate to C. guianensis management guid e lines Implications for MultiUse M anagement Results from this study indicate harvests of timber and NTFPs may be complementary in some situations, even for the same species. Significant income pulses from periodic timber harvest s of C. guianensis could complement a more steady revenue stream from annual seed harvests all of which could be integrated with harvest of other timber and NTFPs. Increasing househ old forest based revenues may

PAGE 87

87 also increase the likelihood that forests will not be converted to other land uses (Barros 2002, Morton et al. 2006). Furthermore, multiple use forest management can lead to increased employment opportunities and diversificat ion of household incomes (Mollinedo 2000, Campos et al. 2001, Mollinedo et al. 2001) without thwarting the central food security role of NTFPs as key subsistence products and buffer s in times of food shortages ( Lipper 2000, Pattanayak and Sills 2001). Man agement guidelines for integrating timber and NTFP harvests will necessarily be different, and perhaps more complex, than those for any one product type. Marking future crop trees prior to logging should integrate not only timber species, but also NTFPs ( Rockwell et al. 2007; Guariguata et al. 2008). Minimizing residual damage may be facilitated in areas where communities or landowners have considerable knowledge of individual trees. If some larger dbh individuals of C. guianensis are found to contribute disproportionately to seed production, as is the case for some species (Snook et al. 2005, Kainer et al. 2007), they could remain unlogged until their seed production diminished beyond some theoretical threshold. Previous research indicated a quadratic relationship between seed production and diameter for C. guianensis in both forest types: smallest and largest size classes rarely contributed to seed production (See Chapter 1). This quadratic relationship has been documented for other species where logging had not previously occurred ( Wadt et al. 2005, but see Snook et al. (2005) and NabeNielsen et al. (2009) for logged examples. This suggests that temporal separation of seed and timber harvests merits consideration. If this quadratic pattern holds across other multi use species (timber and fruits/seeds), harvest of the largest individuals may have a relatively small impact on future population dynamics

PAGE 88

88 though our elasticity results indicate the importance of maintaining high survival for nonlogged indivi duals However, because the largest and oldest individuals of a population tend to positively correlate with size and number of hollow stems (Lindenmayer et al. 1993, 2000, Fox et al. 2008), targeted logging of these top dbh individuals may not be economic ally viable, although incidence of heart rot varies by species (Fox et al. 2008). In contrast to this temporal separation of timber seed harvests, spatially setting aside areas with high densities of good seed producing trees would eliminate potential logging damage. In areas where desirable timber trees are mixed with good seed producers and delaying timber extraction is not economically viable, reducedimpact logging techniques such as vinecutting and directional felling can be used to mitigate damage (P inard et al. 1995, Dykstra & Heinrich 1996). Conversely, if careful analysis indicates that an individual is more valuable for timber than for its seed production, logging may be an appropriate decision. Tropical forest conservation increasingly relies on multiple products and services valued in rapidly changing markets and landscapes. C. guianensis is one potential species for such multi use management. Our results indicate potential compatibility of timber and seed harvests in occasionally inundated forests of Acre, Brazil. Further research to determine whether these results hold for other economically important tropical species could increase the diversity of management options for Amazonia.

PAGE 89

89 CHAPTER 4 ECONOMIC REVENUE FRO M SUSTAINABLE SEED AND TIMBER HA RVESTS OF CARAPA GUIANENSIS Introduction Integrating different revenue sources within management regimes may be the future of tropical forest management. While logging is currently one of the main forest uses in the Amazon due to its importance in the reg ional economy and the high present value associated with singleentry timber extraction (Verssimo et al. 1992, Lentini et al. 2003, AIMEX 2005), it is also one of the most intensive and extensive sources of forest management. While the principle of presen t value maximization is the driving force that determines land use in a market economy (Klemperer 1996), indicating that timber extraction will likely retain its importance in tropical forest management (Pearce et al. 2003), logging has significant drawbac ks: it removes the entire individual, requiring decades for stem replacement given growth rates, incurs significant extraction costs (Holmes et al. 2002), can increase forest flammability (Nepstad et al. 1999), and encourages conversion of forest to other land use (Asner et al. 2005) Management of nontimber forest products has also been touted as an economically viable method of forest management. NTFP harvesting can contribute to the economic livelihoods and welfare of forest residents, can result in less ecological destruction than timber harvesting and other forest uses (Ticktin 2004), and may add to the perceived value of standing tropical forest (Chopra 1993, Gunatilake et al. 1993, Marshall et al. 2003). However, NTFP overharvesting does occur (Gaou e & Ticktin 2009), causing alteration or degradation to the resource and affecting species persistence (Peres et. al. 2003, Ticktin 2004). NTFP harvests also may not be sufficient to lift communities out of poverty (Morsello 2006).

PAGE 90

90 If linked, however, mult iple use forest management could provide the best of both worlds, pooling the income periodically available from timber harvest within a backdrop of steady income from NTFP harvest. Proponents of diversified forest management highlight the f act that integr ating multiple forest values provides a social and financial edge in comparison with timber dominant management models (Panayotou & Ashton 1993, Salick et al. 1995, Scherr et al. 2003). Indeed, multipleuse management of tropical forests (MTF) is not a new concept (Panayotou & Ashton 1992, Salick et al. 1995), but rather is the standard in most smallholder and community managed forests (Whitmore 1990). The combined extraction of xate ( Chamaedorea spp,) and timber in community forests in Guatemala is one suc cessful example of MTF (Guariguata et al. 2008); management for chicle, honey, game and timber in Mexico is another community managed MTF regime acknowledged for its success (Snook 2000). There are, however, still multiple ecological, social, and economic barriers to widespread successful implementation of MTF (GarcaFernndez et al. 2008, Ros Tonen et al. 2008). While there is a growing body of literature on the economics of timber harvest (Boltz et al. 2001, Merry et al. 2009) as well as on economic ret urns from individual NTFPs (Varghese & Ticktin 2008, AvocvouAyisso et al. 2009) there is still limited research on the compatibility of timber with nontimber forest management (though see Snook 2000, Guariguata et al. 2008, Menton et al. 2009, Guariguata et al. 2010). Most studies have have highlighted the ecological benefits of not just focusing on timber without incorporating the economic opportunities and tradeoffs associated with multiple use management (though see Boscolo & Vincent 2003, Menton et al. 2009). Quantifying

PAGE 91

91 revenue from harvest is particularly important in cases where an individual species provides multiple economic benefits, generating a potential conflict of interest in deciding whether to prioritize one use over another. Such conf lict of interest is common in the Amazon: HerreroJuregui et al. (2009) found that 46% of the 200 timber species in the state of Par, Brazil, also provided nontimber forest products and reportedly at least one third of 305 Amazonian timber speci es als o provide NTFPs (Martini et al. 1994). Multiple use species should be managed to maximize their economic or social value under constraints of ecological sustainability. The determination of how to do this is often left to the forest resident or manager who frequently uses imperfect and general information on the benefits of different forest management strategies to make a decision that meets particular objectives (Macpherson 2007). Even if detailed knowledge of sustainability does not always translate into sustainable management practices (Gaoue & Ticktin 2009) and expected economic gains are not always realized (Putz et al. 2000), improving the information available to managers should enhance both ecological and economic decisionmaking. This study augments the economic information available to forest managers, in particular the potential revenue associated with different harvest strategies of the multipurpose tropical tree Carapa guianensis C. guianensis is valued for both the high quality oil extracted from its seeds (Shanley & Medina 2005) and its mahogany like timber (McHargue & Hartsorn 1983, Mabberley 1987). Pure C. guianensis seed oil is used for medicinal applications (Rodrigues 1989), with valueadded products including soaps, shampoos, candles and repellent torches (Shanley & Medina 2005). This

PAGE 92

92 species is considered to have such great economic potential that the Amazonian State of Acre in Brazil has identified it as one of six priority species for extraction research (Acre 2000). C. guianensis timber and seed harvest s can be mutually compatible: a quadratic relationship between size and seed production found in previous research (See Chapter 2) coupled with relatively low numbers of individuals in the larger harvest classes indicated that in some cases, larger individuals could be harvested for timber in conjunction with low levels of seed harvest while maintaining positive population growth ( See Chapter 3). Other species also demonstrate this quadratic relationship between size and seed production (Soehartono & Newton 2001, Kainer et al. 2007); in these cases, the largest individuals could potentially be harvested for timber, leaving midsized individuals for seed production. A continuum of options may be more appropriate, however, especially for communities that have knowledge of individual trees. For example, trees that have consistent low seed production or are past their reproductive prime may be suitable candidates for selective logging. Spatial segregation of timber and NTFP management unit s is another approach for concurrent timber and NTFP management (Guariguata et al. 2010). A comparison of revenue generated by individual tree harvest and seed collection would allow forest managers to better choose a management strategy that meets their economic objectives. W e provide one such comparison using C. guainensis modeling the equal annual equivalent (EAE), a function of net present value (NPV), of revenues associated with modeled simulations of sustainable seed and timber harvests. Our specifi c objectives were 1) to simulate and compare the revenue from ecologically

PAGE 93

93 viable seed and timber harvests of C. guianensis in stochastically varying environments; and 2) to simulate the EAE of revenues from seed and timber harvest under different market prices for both seed oil and timber. Sustainable harvest simulations are based on a stochastic ecological matrix model parameterized with data from a demographic study in the southwestern Amazonian state of Acre, Brazil (See Chapter 3). Here, we only provide a brief, summative explanation of this detailed ecological model necessary to understand the integration of ecological parameters within the context of our economics model. Methods Study Site and Field M easurements We carried out field measurements within the 1200 ha experimental forest of the Brazilian Agricultural Research Corporation (Embrapa) in the eastern portion of the state of Acre ( latitude 10 01' 28" S and longitude 67 42' 19" W). The region has rolling topography including occasionally inund ated areas and upland habitat ( terra firme ). Two 400 m x 400 m (16ha) plots were established where the majority of the environment was classified as terra firme forest and two in occasionally inundated forest (Klimas et al. 2007). We inventoried and monit ored survival of all individuals > 10 cm diameter at breast height ( dbh ) from 2004 to 2009. We also recorded canopy position (dominant, co dominant, intermediate or suppressed), spatial coordinates, and initial diameter at breast height (dbh) for each tree (measured in 2004) We installed dendrometer bands on a random subset of over 500 trees stratified by dbh and forest type to measure annual growth. Using this subset we also measured crosssectional crown area, tree height, tree commercial height and tre e form. We measured crown cross sectional area along two axes: maximum crown diameter and a second diameter formed at a right

PAGE 94

94 angle to the former, and subsequently applied an ellipsoid formula to convert axis diameters to crown area (Kainer et al. 2007). T ree heights were measured with a S uunto optical height meter PM 5/1520 (clinometer), using a surveyors tape to determine distance from clinometer to tree. We recorded both total and commercial tree height; commercial height was measured to the highest def ect free point on the tree trunk and excluded if tree form invalidated commercial h arvest We made notations of any abnormalities that affected trunk form, such as evidence of a previous burn, forked trunks (a potential indication o f prior insect attack), and damaged or absent crowns Rainfall and temperature data were collected daily at noon at the Federal University of Acre (UFAC), approximately 8 km from our study site. Average annual temperature was 24.5C with a dry season from June to August. Monthly mean temperature was averaged from the daily compensated mean temperatures for a given month. Maximum and minimum temperatures also represented monthly averages of these daily temperature values (INMET 2009). Seed Oil Extraction and Price E stimates We c onducted interviews with key informants for market data on the price received for C. guianensis seed oil in 3 locations in the state of Acre from December to February of 2009. We utilized snowball sampling (Vogt 2005): initial interviews led to referrals t hat generate d more interviews with individuals or companies engaged in either purchase of seed oil or in oil extraction. We visited different areas throughout the state where individuals or cooperatives ex tracted seed oil to estimate yield in kilograms o f seeds needed to make 1L of C. guianensis seed oil, the dominant marketable unit of oil We also noted any procedural

PAGE 95

95 differences in oil extraction between sites. We use d a prior study to calculate labor costs associated with oil extraction (Londres 2004) Timber V alue We used two sets of allometric volume equations to determine stem volume of harvested C. guianensis individuals which we then multiplied by price/ m3 and the number of trees available for harvest. The first equation of tree volume was developed by Segura & Kanninen (2005) based on measurements of 19 trees destructively sampled trees in Northern Costa Rica, including 3 C. guianensis individuals Their best fit model for stem volume was: Vstem = exp( c + a ln DBH) where c = 8.70 1.77 and a = 2.41 0.40 (Segura & Kanninen 2005). These models, however, were constructed using indivi duals with diameters between 60 and 105 cm, and tended to overestimate tree biomass and volume outside the specified diameter range. Thus, we also used a sec ond set of volume equations specific to C. guianensis from an Amazonian data set of measured dbh and commercial tree height s from Par, Brazil (Silva et al. 1984). Equations that include both commercial height and dbh are normally better at predicting the actual volume, though commercial tree height can be difficult to measure with accuracy ( Segura & Kanninen 2005). Silva et al. (1984) used 183 C. guianensis sample trees to find co nstants most appropriate for volume equations, and parameterization did not i ncl ude destructive harvest s (Silva et al. 1984). The best fit model for Carapa tree volume (without bark) was: V = exp( 9.20315 + 2.01914 ln d + 0.76028 ln h )

PAGE 96

96 where V is the stem volume without bark, d is the dbh, and h is the commercial height excludi ng stump ; these values did not include loss es during log processing (Silva et al. 1984). We used a dbh of 50 cm ( d ) and the average height of sampled individuals with dbh h ) to solve for timber harvest volume ; a dbh of 40 cm was used to calculate illegal harvests of individuals ranging in diameter from 40 to 50 cm While using this lower dbh limit may underestimate initial harvest of typic ally larger diameter trees, it should be more representative of future harvests. Slow growth rates mean that mo st individuals will be harvested soon after they reach harvestable diameter Indeed, C. guianensis individuals spend an average of 20 years to grow from 40 to 50 cm dbh and from 50 to 60 cm, but the standard deviation of growth was high (55 years). We thin k that these slow growth rates, coupled with high variability, warrant conservative estimates of return from timber harvests. Finally, to determine the current timber market price, we interviewed researchers at Embrapa specializing in timber extraction. Su stainable Harvest Simulations Under Simulated Environmental S tochasticity This study builds on a previously developed stochastic ecological simulation model developed in MATLAB (2002) We simulated s tochastic environmental conditions using the independent and identically distributed environmental states represented by four transition matrices ( See Chapter 3 for additional information on model construction). We assumed that the four years of our study represented the range of typical environmental conditions each of which was equall y likely to occur in the future. E ach modeled year 1 of 4 transition matrices was chosen at random and used to simulate population growth. To explore possible sustainable harvesting regimes in the simulation model, we reduced

PAGE 97

97 ave rage fecundity (to simulate seed harvest) and the number of individuals in the largest size classes (to simulate timber harvest) ( See Chapter 3). These stochastic simulation model results indicated compatibility of seed and timber harvests in occasionall y inundated forest under certain harvest levels Harvests of 10% of annual seed production combined wi th 100% timber harvests every 25 years maintained the simulated stochastic population growth rate above one s = 1.0012) indicating an increasing population, though the lower 95% confidence interval of s fell below one (0.9991, 1.0033) These same har vest levels simulated a decreasing population in terra firme s = 0.9663), but since population dynamics are not normally considered in harvest decisions, we also calculated revenue from these harvest scenarios. Since the stochastic model used sizeclass specific estimates of seedling fecundity (average number of seedlings produced by an individual in a given dbh class), we associated the model with estimates of average seed production per individual in a given dbh class based on quantification of five year s of seed production in the study area ( See Chapter 2). Timber h arvest scenarios included both legal and illegal timber harvests Only individuals dbh are legally harvestable based on current Amazonian forest law ( Resoluo N. 406) and legal cutting cycles vary from 25to 35 year s in Amazonia (Instruo Normativa 05/2006, Resoluo N. 406) So me authors state that this shorter cycle is likely not sustainable (Gardingen et al. 2006) and C. guianensis is a slow growing species (Vieira et al. 2005 Chapter 3). Indeed our model results indicated that timber harvests of the largest individuals ( cm dbh) were not sustainable in terra firme forests, though harvests were arguably sustainable in occasionally inundated

PAGE 98

98 forest (but the lower 95% confidence interval indicated potential for nonsustainable harvest). Net Present Value of Seed and T imber H arvests To calculate seed harvest revenues, the results indicating ecologically sustainable seed harvest levels from MATLAB matrix model results were converted to estimates of oil yield and multiplied by different observed market prices to generate estim ates of expected revenue from yearly seed harvests. For timber harvests, the number of individuals harvested at 50year intervals was multiplied by the volume estimates from the two allometric equations and then multiplied by the market price of a cubic meter of C. guianensis timber. The present value (PV) of all future harvests was calculated using the equation: PV = Vh*e( r*t) where Vh is the yearly harvest value, r is the discount rate and t is the year Revenues from future seed harvests and logging entries were discounted at 4%, 8% and 20% to examine the impact of discounting on the profitability of long term management. We calculated the net present value of annual seed harvests and both 25 and 35year timber harvests based on simulations ; these NPVs were then converted to a per hectare EAE (see below) The NPV of each years seed harvest was stored as a vector and summed over the 5171 years of the model (the temporal period was based on 3timber cutting cycles) to determine the NPV of seed harvest over that same time period. We deducted labor costs per kilogram of seed oil processed based on an estimated daily wage of R$23 (the minimum salary in Acre for agricultural workers, equal to US$ 12.88 on 7/19/ 2010) B ased on Londres ( 2004), two full days of labor yields two kilograms of oil (or 1 day per 1 kg). Londres estimate included 4 hours for cooking

PAGE 99

99 seeds, 4 hours for breaking the m and removing the endosperm (30 days later), and 15 hours to prepare the seed pulp for oil extraction and mix for 15 minut es 3 times per day (for 15 days) Seed collection was combined with other activities (Londres 2004) and was not assigned a time cost. In addition to using a labor cost of R$23 (US$12.88) w e also calculated revenue using a labor cost of R$12 (US$6.72 on 7/ 19/2010) per kilogram of oil to assess how sensitive revenue was to labor cost. We used a similar methodology to calculate the NPV of our 25and 35 year timber harvests. We simulated and summed the NPV harvests 500 times to create a NPV distribution for both seed and timber harvests; these simulations represent how ecological variability may influence expected revenue. We used these NPV sums to calculate the equal annual equivalent (E AE). The EAE allows comparison of land uses that have different investme nt time periods (annual vs. periodic harvests) by combining all costs and benefits into a single annual sum that is the equivalent to all cash flows during an analysis period spread uniformly over the period (Jacobson 1998). EAE = NPV 1 ) 1 ( ) 1 ( t ti i i where i is the discount rate, and t is the number of years. The EAE allowed for comparison of different cutting cycles (i.e. 35year cutting cycles vs. 25year cutting cycles). All EAE results are expressed on a revenue per hectare basis. Results Se ed Oil V alue Prices for C. guianensis seed oil in 2009 varied throughout the state of Acre. A soapmaking company, Tawaya, in the states most western city of Cruzeiro do Sul paid

PAGE 100

100 R$15/kilogram (US$8.40 on 7/19/2010) of seed oil. A community cooperative, RECA in the neighboring state of Rondnia, sells C. guianensis se ed oil for R$25/kilogram (US$14 on 7/19/2010) and published values for th e northern Amazon were R$3 (US$1.68 on 7/19/2010) in 1999 (Plowden 2004). Oil extraction methods observed at both the community cooperative in Rondnia and in Cruzeiro do Sul used dried seeds and yielded comparable kilograms of oil In Rondnia, one kilogram of seeds yielded 0.432 ( 0.036 SD) kilograms of unfiltered oil, using a mechanical steel press after 30 minutes. Using agricultural equipment modified to break and press seeds and extract oil in Cruzero do Sul, one kg of seeds yielded 0.450 kg of oil after pressing for one ho ur and 40 minutes Manual extraction results in comparable oil yields, but time for extrac tion is much greater than that using the modified agricultural equipment (RECA communication). We chose to use the estimate of oil yield from Rondnia (0.432 kilograms of oil per 1 kilogram seeds) in our seed harvest NPV calculations since it was comparabl e to manual extraction yield and provides a conservative estimate of oil yield. Because labor costs (time expended) were highly variable and poorly quantified in these sites, we instead adopted detailed estimates from Londres (2004) as outlined in our met hods. Timber V alue The local stumpage price of a standing C. guianensis tree was R$40 to R50/m3 (US$ 22.40 to US$28 on 7/19/2010) Commercial heights of individuals ranged from 8.5 to 20 m, with a tendency toward lower heig hts, and average commercial height of all individuals e timber volume. Prices for extracted C. guianensis stems were R$150/m3 on a log deck (US$83.99 on 7/19/2010) and between R$600 R$ 900/m3 (US$335.95$503.92 on 7/19/2010) when processed

PAGE 101

101 a t sawmills in the capital city of Acre, Rio Branco (with a 3:1 yield ratio) ( Figueiredo, personal communication). Since most communities in Acre do not have the training or equipment necessary to extract timber, or process at onsite sawmills, we used stum page price to calculate timber revenues Net Present Value of Seed and Timber H arvests W e found that revenue from 100% timber harvests (of individuals 25years) exceeded that of 10% seed harvests when evaluated separately and applying labor costs of R$23 per kilogram of oil (US$12.88 on 7/19/2010) When we dropped those l abor costs to R$12 (US$6.72 on 7/19/2010) however, e xpected seed harvest revenue jumped by a factor of 6 or more, surpassing the EAE of timber harvest (Table 4 1). These trends also were consistent for 35year cutting cycles (results not shown), though EAE of both seed and timber harvests for this longer cutting cycle were less t han in simulated 25year cycles. Revenue patterns were consistent in both occasionally inundated and terra firme forests, though expected revenue from bo th seed and timber harvests was consistently higher in occasionally inundated forests (Table 41; Table 42) The obvious caveat here is that timber extraction only included C. guianensis Most selective logging is not exclusive to one species, but rather includes a mix of valuable timber species.

PAGE 102

102 Table 41 Expected revenue per hect are of C. guianensis seed and timber harvest in an occasionally inundated forest (A) and terra firme forest (B) in Acre, Brazil. All revenues indicate the total revenue over 51years or 3 timber cutting cycles (mean, minimum and maximum values are from 500 simulations ). The seed price is bas ed on the market value of 1 kg of processed seed oil; timber price is based on the value of a m3 of timber. Prices and revenues are expressed in the Brazilian real (plural: reais ) currency ( 1 real = US$0.56 7/19/2010). F or timber, the first revenue listed is calculated using multi species allometric equations derived from Costa Rica n sites (Segura & Kanninen 2005) The second timber revenues in parentheses are based on allometric equations of C. guianensis ( Silva et al. 1 984) Revenues for a 20% discount rate were approximately zero and are not shown. A: Occasionally inundated forest Seeds Timber Discount rate Harvest Labor (R$) Price (R$) Simulated EAE Range (R$) Mean Min Max Price (R$) Simulated EAE Range (R$) Mean Min Max 4% 10% seeds and 100% timber 23 25 0.29 0.13 0.46 40 0.79 (0.83) 0.73 (0.76) 0.83 (0.89) 12 25 1.90 1.07 2.92 100% illegal timber harvest 23 25 40 1.99 (2.11) 1.89 (2.02) 2.07 (2.22) 12 25 8% 10% seeds & 100% timber 23 25 0.03 0.01 0.06 40 0.10 0. 10 0. 10 12 25 0.22 0.09 0.34 100% illegal timber harvest 23 25 0.30 0.29 0.31 12 25

PAGE 103

103 Table 41. Continued B: Terra firme forest Seeds Timber Discount rat e Harvest Labor (R$) Price (R$) Simulated EAE Range (R$) Mean Min Max Price (R$) Simulated EAE Range (R$) Mean Min Max 4% 10% seeds and 100% timber 23 25 0.08 0.03 0.13 40 0.32 (0.34) 0.29 (0.31) 0.35 (0.37) 12 25 0.52 0.25 0.86 100% illegal timber harvest 23 25 40 0.73 (0.77) 0.67 (0.70) 0.79 (0.86) 12 25 8% 10% seeds & 100% timber 23 25 0.01 0.00 0.02 40 0.05 0.05 0.05 12 25 0.06 0.02 0.11 100% illegal timber harvest 23 25 0.12 0.11 0.12 12 25

PAGE 104

104 Table 42 Expected revenue per hectare of C. guianensis seed and timber harvest in an occasionally inundated and terra firme forest in Acre, Brazil. All revenues indicate the total revenue over 51years or 3 timber cutting cycles (mean, minimum and maximum values are from 500 simulations). The seed price is based on the market value of 1 kilo of processed seed oil using R$25 (US$14 on 7/19/2010) marke t price and a labor cost of R$23 (US$12.88) or R$12 (US$6.72) ; ti mber pric e is based on a stumpage value of R$40 (US$22.40) per m3 of timber. Prices and revenues are expressed in the Brazilian real (plural: reais) currency (1 real = US$0.56 7/19 /2010). Timber revenue estimates were calculated using Segura and Kanninen s (2005) m ultispecies allometric equations from Costa Rican sites Revenues for a 20% discount rate were approximately zero and are not shown. Occasionally inundated forest Terra firme forest Discount rate Harvest Labor (R$) Simulated EAE Range (R$) Mean Mi n Max Simulated EAE Range (R$) Mean Min M ax 4% 10% seeds and 100% timber 23 1.07 0.95 1.19 0.40 0.33 0.47 12 2.68 1.90 3.66 0.84 0.54 1.20 100% illegal timber harvest 1.99 1.89 2.07 0.73 0.67 0.79 8% 10% seeds & 100% t imber 23 0.13 0.11 0.15 0.06 0.05 0.07 12 0.32 0.20 0.43 0.11 0.07 0.17 100% illegal timber harvest 0.30 0.29 0.31 0.12 0.11 0.12 Timber and seed extraction, however were not mutually exc lusive at some levels modeled (10% seeds and 100% timber). These combined harvest levels had a s greater than one for occasionally inundated forests (See Chapter 3) indicating ecologically viable harvests, and positive revenu e in each forest type (Table 42 ) Annual oil harvests varied considerably between years and between forest types due to annual variability in seed production (Figure 4 1 ) In contrast, timber harvests showed relatively little variation due to tree slow annual growth increment The number of individuals that grew to harvestable size within the 25year cutting cycles was low, r eflecting, harvest intensities of less than three individuals per hectare, even with harvests of all individuals dbh 40 cm (Table 43) Based on distributions of NPVs for

PAGE 105

105 10% seed harvest predicted from 500 simulations (Figure 4 2), w e observed an expected decrease in NPV wit h increasing discount rate. We also observed ecological variability as seen in the variation ar ound mean NPV; as discount rate increased this ecological variability had less impact on expected revenue. A B Figure 41 Potential annual seed oil production per hectare with harvests of 10% and 20% of the seed crop in occasionally inundated (A) and t erra firme forest (B). Variability is due to modeled variability in seed production between years.

PAGE 106

106 Figure 42 Histograms of the dist ributions of NPVs for 10% seed harvest predicted from 500 simulations using a market price of R$25 and a labor cost of R $12 ; variability is due to differences in expected seed production based on modeled ecological stochasticity. Table 43 Average number of individuals harvested per hectare for timber based on a 25year cutting cycle. Harvest Occasionally inundated for est Terra firme forest Number of individuals per hectare Year dbh 50 cm dbh 40 cm dbh 50 cm dbh 40 cm 1 0.87 2.97 0.53 1.27 26 1.76 2.99 0.41 1.10 51 1.62 2.61 0.26 0.45

PAGE 107

107 Discussion When identifying potential management regime s for tropical forests, civil society prefer s activities that are biologically feasible, ecologically sustainable, economically efficient, socially acceptable and operationally feasible. While very few alternatives pass all these tests (Alavalapati & Zarin 2004), our study found that both 10% seed harvests and 100% timber harvests of C guianensis (individuals dbh 50 cm) were both profitable and arguably ecologically sustainable, though in occasionally inundated forests only Timber harvest of C. guianensis was economically superior to income from seed harvest when the full cost of seed oil extraction was included, but when labor costs were halved, revenues from seed harvests were competitive or superior to timber harvests (Table 41). Reducing labor costs was based on the assumption that individuals involved in the seed processing were not always earning the equiv alent of a daily minimum wage, therefore we may have overestimated the true opportunity cost of their labor. In practice, however, benefit s could accrue from both annual seed harvest revenues and those from periodic logging revenues of C. guianensis indiv iduals. Shadow P rices Labor costs differed between the two C. guianensis products. For modeling, we assumed no labor costs for timber extraction because timber prices modeled were based on stumpage values (standing trees harvested by loggers) Our modeled labor costs for seed collection and oil extraction (R$23 per kilo of oil extracted) may have overestimated the true cost of labor due to shadow prices. A shadow price differs from a financial price in that it reflects the true opportunity cost of the reso urce in question. While oil extraction takes 25 months without the use of mechanized equipment ( Londres 2004, Plowden 2004), the associated work is not continuous because of

PAGE 108

108 significant time lapses as oil drains during different processing stages. Thus, parts of the oil extraction process can be combined with other household activities If, however, communities plan on increasing seed harvests to above 20% of total seed production simulated here, seed collection and processing would require substantial ly g reater amounts of time. The costs of extensive seed collection are not included in this study due to the expectation that market demand is not sufficient to warrant this scenario. Transport costs and potential capital investment also differed between C. guianensis timber and seeds. For timber, t ransport costs to market and capital expenditures are inc luded in our estimates of stumpage price s. We also assumed no transport costs for marketing seed oil since this was the case for individuals involved in smallscale oil extraction in Acre at the time of our study (personal obser vation). Community members travelled to urban centers on a relatively frequent basis to visit relatives and conduct other transactions. When marketing of seed oil was combined with these pre planned t rips, the opportunity cost was effectively zero. Because C. guianensis oil can be stored after processing (RECA, personal communication), often in repurposed cont ainers such as 1 L soda bottles, we did not subtract any oil transport costs fro m our expected net present value of seed harvest. If, however, communities were to invest in larger scale extraction operations, a more thorough analysis of initial capital investment costs would be necessary (including receptacles for oil storage and tran sport). Model A ssumptions This model only incorporates logging of one species at very low densities This scenario is limited in scope; most logging currently includes other valuable timber species identified in preharvest inventories. Additionall y, we ma y have overestimated

PAGE 109

109 the expected revenue from timber harvest since we did not include mortality from tree felling in our model. Holmes et al. (2002), however, only estimated additional mortality at 2.6% when multiple timber species are harvested using log ging practices designed to reduce negative environmental impacts on a given stand. Our simulated harvest intensities were low, including the equivalent of 1 3 in dividuals per hectare at each 25year interval (Table 43) Still, addit ional mortality could a ffect stasis (survival) of this species, one of the most important demographic factors in predicting C. guianensis population growth rates (See Chapter 3). This omission, however, is perhaps compensated for by the fact that we do not assume higher growth ( due to increases in light) immediately after timber harvests as is common in some logging models (Boltz et al. 2001, Valle et al. 2007 ). As our logging intensities are low, increased growth would affect only a few individuals. Increased growth is also reportedly short lived, both in the Brazilian Amazon (Silva et al. 1995) and Surinam (De Graaf et al. 1999, Dekker & De Graaf 2003). Limitations of Model R esults Population stabili ty is a precondition to sustainable revenues over time. As evidenced by our ti mber simulations in terra firme timber extraction decreased over the interval of just 3cutting cycles (Table 43) Simulating increased timber harvests cannot be justified, since C. guianensis growth rates were some of the slowest in the Amazon (Vieira e t al. 2005). Revenues from timber harvests could potentially be higher with increased processing, however. Prices of C. guianensis timber jumps R$100 just by removing it from the forest to a log deck; this, however, requires labor, training, and an associa ted time cost.

PAGE 110

110 In terms of C. guianensis seeds, harvest s may be higher in other Amazonian regions where hi gher average seed production has been reported (Tonini et al. 2008, Guedes et al. 2008, Londres 2009), al though their production estimates were based on fewer years of observations and C. guianensis shows pronounced variation in seed production between years ( See Chapter 2). The market for C. guianensis seed oil is also moredeveloped in the Northeastern Brazilian Amazon; these markets might be able to absorb higher oil production than smaller markets in the state of Acre. It is important, however, to exercise caution when hypothesizing about prospective gains from increased product extraction. Local and regional markets cannot always absorb the volume o f products that could result from highly intensified extraction ( Padoch 1988, Padoch & PinedoVasquez 1996, Shanley et al. 2002). Best Management Practices for MultipleU se C. guianensis M anagement Maximizing stumpage value while achieving a minimum level of nontimber benefits likely requires harvesting rules that are more complex than the simple diameter limit system implemented in most parts of the Amazon and used in this study ( Boscolo & Buongiorno 1997, Buongiorno et al. 1995 ). Designing these systems requires a great deal of silvicultural expertise as well as the capacity to monitor logging operations. Monitoring may be easier to implement in communities that hav e a good knowledge of diverse tree species (and perhaps individual trees) and a low opportunity cost of labor (Boscolo & Vincent 2003). Spatial separation of management units, or specialized management, where some stands are managed more intensely for timber and other stands more intensely for NTFPs, is one management option (Binkley 1997,da Sil va Dias et al. 2002, Zhang 2005, Guariguata et al. 2010). Boscolo & Vincent (2003) found that specialized management might be superior for combined biodiversity conservation

PAGE 111

111 and production of timber at low discount rates. Specialized management units might be useful for C. guianensis timber and NTFP combined management as well. Communities could identify those trees that should remain unlogged for their value as seedproducers either because they are not fit for timber or are consistently high seed producers. Research on Bertholletia excelsa (Kainer et al. 2007) and Swietenia macrophylla ( Snook et al. 2005) have documented the latter, while also noting individuals that produce few to no seeds which could be targeted for logging. Field observations of C. gui anensis in our study site also indicated a subset of trees with consistently lower seed production. Planning timber management to target these marked low seed producers is one way to minimize impact of timber harvest on the seed resource. This requires inf ormed planning and organization, which are not consistently found in Amazonian communities (Shanley et al. 2002) though this may be changing. I nitiatives are underway in Brazil to train tropical foresters in bridging the gap between timber and NTFP use, ecology and management (Pinto et al. 2008, Shanley & Medina 2005). Managing timber resources for continued extraction (as opposed to depletion harvesting) has a short history in the Amazon basin; there is no experience beyond the first harvest (Zarin et al. 2007). In selective logging, first harvests yield high timber volumes because they take place in forests that have not been anthropogenically disturbed for hundreds of years. Second cuts are not nearly as profitable (Keller et al. 2007, Macpherson 2007), and concerns about viability and returns associated with future timber harvests can lead to land use conversion for cattle ranching and slashandburn agriculture (Verssimo et al. 1992, Nepstad et al. 2001) in the absence of other

PAGE 112

112 economic incentives from forest management between cutting cycles. Our combined returns for seed and timber harvest of C. guianensis indicate a potentially viable economic opportunity that is sustainable in certain forest types. Combining extraction of NTFPs with timber may provi de the needed incentive for maintaining forest cover until regrowth is sufficient for the next timber harvest. This is even more important if the current timber cutting cycles are unsustainable ( Gardingen et al. 2006) and therefore adjusted upwards, further reducing the expected value from future harvests of timber only. This study found significant estimated revenue from both timber and seed harvest of C. guianensis though this revenue is not likely sufficient to move communities out of poverty Combining this income with harvest of other resources in the context of multiuse management may be an economically viable management strategy in Amazonia preferable to some current singleresource extraction management systems.

PAGE 113

113 CHAPTER 5 CONCLUSION Taken togeth er, the three papers in this dissertation allow for a better understanding of the ecological and environmental mechanisms that influence C. guianensis natural seed production and variability, how the ecology of the species constrains both seed and timber harvests in two different forest types, and the potential economic revenue from simulated ecologically sustainable harvests. The first paper demonstrated variable seed production between years and between individuals Multiple variables interact ed at differ ent scales to influence seed production. At the largest scale, climatic cues (rainfall and temperature parameters) were central to setting overall patterns in phenophases and seemed to best explain why high seed production years were consistent across bot h forest types (or habitats) examined. Habitat heterogeneity, including species abundance, spacing between individuals (Klimas et al. 2007) and underlying abiotic factors may be the primary reason for significantly higher seed production in occasionally inundated forest (F = 22.60, Table 2 2), though climatic variables had a much stronger predictor effect (Table 22). At the smallest scale, individual tree attributes contributed to seed production heterogeneity within habitats; dbh, and cross sectional ca nopy area showed positive, quadratic relationships with seed production while vine load negatively affected seed production, irrespective of forest type. Mean annual maximum temperature was negatively correlated with C. guianensis seed production perha ps due to its often strong correlation with irradiance (Van Schaik et al. 1993, Stevenson et al. 2008). Threemonth dry season rainfall was positively correlated with seed production, while fivemonth wet season rainfall was negatively

PAGE 114

114 correlated. Higher t han normal dry season rains may allow for increased available reserves for fruiting and flowering. Too much rain, however, may inhibit synchronization of flowering ( Alvim 1960 ) and reduce pollinator visits ( Van Schaik et al. 1993). C. guianensis had signif icantly higher seed production in occasionally inundated forest with higher densities in this forest type further amplifying production. The interaction between forest type and year was also significant in our peak seedfall model (Table 23). Perhaps part of this interaction and likely the observed lower seed production in terra firme forest, can be explained by a 2005 fire in one of our terra firme plots which stressed trees and reduced stored reserves available for reproduction. This fire, however, c oincided with an especially severe Amazonian drought that would have affected both forest types. Indeed, percentage of individuals flowering and total seed production were low in both forest types in 2005 and the subsequent year Still, our study cannot confirm this fire hypothesis since we do not have data prior to the drought The second paper focused on sustainable harvest scenarios for C. guianensis Results indicated that some populations of C. guianensis may be suitable for extraction of seeds, timber or a combination of both. No levels of timber or seed harvest were considered viable in terra firme s < 1, though a major cav eat of our study is that we measured only 4 transition matrices for a species that can live hundreds of years (Vie ira et al. 2005). If the observed environmental conditions were characteristic of expected future conditions, we would expect a decline in the terra firme population. Indeed quasi extinction simulations indicated that there was a 100% probability that the population would have less than 3 individuals dbh

PAGE 115

115 even in the absence of seed or timber harvest s. Multiple harvesting options, however, were available to sustainably manage C. guianensis in occasionally inundated forest. Simulations predicted population growth in occasionally inundated forest with harvest regimes of 10% and 20% of total seed production, but the lower 95% CI was below one for both of these harvest scenarios (Table 35). Timber harvests of both 50% and 100 % of individuals with dbh the population would have fewer than 3 individuals dbh 10 cm remained zero for the 500 years modeled. W e found this surprising since stasis had some of the highest elasticity values; however, modeled harvest levels were extremely low, involving fewer than 2 trees per hectare every 25years in occasionally inundated forests. Since smaller individuals (< 50 cm dbh) comprised a larger part of the per hectare densities, simulat ions did not reduce the population to fewer than 3 individuals dbh 10 cm per hectare until almost 1000 years in the future. Simulations that limited recruitment, while simultaneously harvesting larger individuals, affected both recruitment of individuals into the smaller dbh classes and seed production in the larger dbh classes, leading to cumulative distribution functions (CDFs) reaching 1.0 after 150 years. We also found the potential for compatibility between seed and timber harvest in occasionally inundated forests. Our results suggested that 10% annual seed harvests were compatible with 100% timber harvests (of trees equivalent of approximately two trees per hectare. The longevity of C. guianensis seen in its high stasis rates, compensated for slow annual growth. Indeed, the importance o f survival (stasis) on population growth rate (based on elasticity calculations) was consistent in both forest types and throughout the 5year study.

PAGE 116

116 The third paper focused on comparing the economic revenue associated with the simulated ecologically sust ainable seed and timber harvests from the previous paper. Quantifying revenue from harvest is particularly important in cases where an individual species provides multiple economic benefits, generating a potential conflict of interest in deciding whether t o prioritize one use over another, as is the case for C. guianensis When we modeled the EAE of expected economic revenue from both seed and timber harvest over a 51year period, we found that revenue from 100% timber harvests (of individuals 25years) exceeded that of 10% seed harvests when evaluated separately and applying labor costs of R$23 per kilogram of oil (US$12.88 on 7/19/2010). When we dropped those labor costs to R$12 (US$6.72 on 7/19/2010), however, expected seed harvest revenue j umped by a factor of 6 or more, surpassing the EAE of timber harvest (Table 41). Reducing labor costs was based on the assumption that individuals involved in the seed processing were not always earning the equivalent of a daily minimum wage; therefore w e may have overestimated the true opportunity cost of their labor. In practice, however, benefits could accrue from both annual seed harvest revenues and those from periodic logging revenues of C. guianensis individuals. A major caveat of this study is th at we simulated timber extraction of C. guianensis soley Most selective logging includes a mix of valuable timber species. Research Significance Few studies have attempted to place an economic value on the flow of goods and services from logging compared with multi use forest management (MTF). This dissertation does not offer a comprehensive comparison of these two strategies when managing an entire forest which remains a challenge (Valle et al. 2007) It does,

PAGE 117

117 however, delve into this issue for one species. This represents an important first step in providing information geared toward encouraging ecologically sound best m anagement practices, while maintaining a profitable harvest scenario. The goal of sus tainable forest management is not to perpetuate p overty in forest based communities, and as such, combining NTFP harvests with other management strategies may be necessary to truly improve livelihoods for forest residents. NTFP harvest alone is currently insufficient (Morsello 2006). Therefore, there is growing interest in incorporating economic analysis with the more common delineation of ecological sustainability of resource harvests to determine whether sustainable harvest levels are expected to be economically viable. Similarly, there is a growing tr end toward analysis of MTF. This includes valuation of diverse forest goods, for example managing for carbon storage in logged stands (Boscolo & Buongiorno 1997, Boscolo & Vincent 2003). Both l ogging and NTFP harvest can be legitimate components of good forest management (Garca Fernndez et al. 2008, Guariguata et al. 2008, Guariguata et al. 2010). In addition, o ne third to one half of timber species are multi use, indicating that they also have value for their NTFPs ( Martini et al. 1994, HerreroJuregui et al. 2009). For MTF to progress, we must combine ecological and economic research to rationally determine the best forest management approaches at the local, regional or national level, including the best management approaches for individual species valued for multiple products. Perhaps incorporating economics with ecological studies can better clarify best management practices under MTF. Evolution of Research and Collaboration I nterests The initial idea for this research, and my interest in forest management, was formed during a oneyear Rotary Ambassadorial Fellowship in Acre, Brazil from 2001-

PAGE 118

118 2002. During this time, I worked as part of a project funded by the International Plant Genetic Research Institute (IPGRI) titled Conservationthroughuse of f orest genetic resources in Latin America. The Acrebased field team worked with four species, one of which was C. guianensis My contribution was predominantly resource mapping of two of these species. Interested in continuing my contribution to research on the viability of forest management as an alternative to forest conversion, I began graduate studies at UF. My graduate project was developed in partnership with Dr. Lcia Wadt, a researcher I had met during my Ambassadorial Fellowship. The project ini tially drew heavily on addressing the question of variable returns from C. guianensis seed harvest due to natural variability in seed production, but grew to include population modeling and economic viability. This research would not have been possible wi thout my collaboration with Dr. Wadt, Embrapa, and the research team we assembled. While I made frequent trips to Brazil, data collection for phenology and seed production was weekly. A team of students and field technicians at Embrapa made this long term data intensive project possible. In turn, I began mentoring/advising two M.S. students at the Federal University of Acre, focusing on statistical assistance in data analysis and helping these students publish in international journals. My partnership wit h Dr. Wadt also enabled me to join a multi site project in Brazil that addressed management of NTFPs in Amazon on a more regional scale. Through continued collaboration with this team, I seek to conduct joint research, plan new projects, and assist in bui lding the capacity of upcoming scientists both in the U.S. and abroad with the general goal of educating

PAGE 119

119 concerned global citizens capable of addressing the major societal and environmental challenges, such as improved forest management in the Amazon. Future R esearch The long term dataset from this dissertation will continue to generate answers to interesting ecological and applied questions. We intend to compare results from this model with results generated via integral projection models, a more sophist icated and arguably more effective modeling technique for modeling the dynamics of long lived plant species (Zuiedema et al. 2010). We also plan to explore in more depth the sources of differences in the long term growth rate between these two forest types under stochastic environmental conditions using recently developed stochastic life table response experiments (Caswell 2010, Davison et al. 2010). Lastly, another hope is procure funding for more applied questions such as the best oil extraction methods a nd oil filtration techniques most appropriate for small communities. The dual economic potential of this species may also lead to increased market research.

PAGE 120

120 LIST OF REFERENCES Ackerman, J.D., Mesler, M.R., Lu, K.L, Montalvo A.M., 1982. Food foragi ng behavior of male Euglossini (Hymeoptera: Apidae): vagabonds or trapliners? Biotropica 14, 241248. Acre, 2000. Governo do Estado do Acre. Programa Estadual de Zoneamento EcolgioEconomico do Acre. Zoneamento ecolgicoeconomico: recursos naturais e mei o ambiente documento final. Rio Branco: SECTMA. V.1. Adams, H.D., Guardiola Claramonte, M. BarronGafford, G.A., Villegas, J.C., Breshears, D.D., Zou, C.B., Troch, P.A., Huxman, T.E. 2009. Temperature sensitivity of drought induced tree mortality portends increased regional dieoff under global changetype drought. Proceedings of the National Academy of Science 106: 70637066. AIMEX, 2005. Estatsticas. Associao das Indstrias Exportadoras de Madeira do Estado do Par. URL: http://www.aimex.com.br Alava lapati, J., Zarin, D.J. 2004. Tropical working forests for what and for whom? Pp. 279289 In Zarin, D.J., Alavalapati, J.R.R., Putz, F.E., Schmink, M. (Eds.) Working forests in the Neotropics: Conservation through sustainable management? New York, Columbia University Press. Alencar, A.A.C., Solorzano, L.A., Nepstad, D.C. 2004. Modeling forest understory fires in an eastern Amazonian landscape. Ecological Applications 14: S139 149. Allen, C.D., Breshears, D.D. 1998. Drought induced shift of a forest woodland ecotone: Rapid landscape response to climate variation. Proceedings of the National Academy of Science U.S.A. 95: 14,83914,842 doi:10.1073/pnas.95.25.14839. Alvim, P. de T., 1960. Moisture stress as a requirement for flowering of coffee. Science. 132, 354. Ares, A., Brauer, D., 2004. Growth and nut production of black walnut in relation to site, tree type and stand conditions in southcentral United States. Agroforestry Systems. 63, 83 90. Ashton, P.S., 1976. Mixed dipterocarp forest and its variation w ith habitat in the Malayan lowlands: a re evaluation at Pasoh. Malayan Forester. 39, 5672. Ashton, P.S., Givnish, T.J., Appanah, S., 1988. Staggered flowering in the Dipterocarpaceae: new insights into floral induction and the evolution of mast fruiting in the aseasonal tropics. American Naturalist. 132, 4466. Asner, G., Knapp, D, Broadbent, E., Oliveira, P., Keller, M., Silva, J. 2005. Selective logging in the Brazilian Amazon. Science 310: 480482.

PAGE 121

121 AvocvouAyisso, C., Sinsin, B., Adgbidi, A., Dossou, G., Van Damme, P. 2009. Sustainable use of non timber forest products: Impact of fruit harvesting on Pentadesma butyrace regeneration and financial analysis of its products trade in Benin. Forest Ecology and Management 257: 19301938. Baillie, I.C., Ashton P.S., Court, M.N., Anderson, J.A.R., Fitzpatrick, E.A., Tinsley, J., 1987. Site characteristics and the distribution of tree species in Mixed Dipterocarp Forest on tertiary sediments in central Sarawak, Malaysia. Journal of Tropical Ecology. 3, 201220. Baker, T.R., Affum Baffoe, K., Burslem, D.F.R.P., Swaine, M.D. 2003. Phenological differences in tree water use and the timing of tropical forest inventories: conclusions from patterns of dry season diameter change. Forest Ecology and Management. 19, 1091 25. Barreto, P., Souza, Jr. C. Anderson, A., Salomo, R. Wiles, J., and Noguern, R. 2005. Human Pressure in the Brazilian Amazon. Instituto do Homen et Meio Ambiente da Amaznia (IMAZON) State of the Amazon 3. IMAZON, Belm, Brazil. Barros, G. 2002. T he economics of beef production in northern Brazil. Indicadores Pecurios, Centro de Estudos Avanados em Economia Aplicada Piracicaba, SP. Belcher, B., Schreckenberg K., 2007. Commercialisation of Nontimber forest products: a reality check. Development Policy Review. 25, 355 377. Bernier, G. 1988. The control of floral evocation and morphogenesis. Annual Review of Plant Physiology and Plant Molecular Biology 39: 175219. Bevan, S.L., North, P.R.J., Grey, W.M.F., Los, O., Plummer, S.E. 2009. Impact of a tmospheric aerosol from biomass burning on Amazon dry season drought. Journal of Geophysical Research 114: D09204, doi:10.1029/2008JD011112. Bhagwat, S.A., Willis, K.J., Birks, H.J.B., Whittaker, R.J. 2008. Agroforestry: a refuge for tropical biodiversity? Trends in Ecology and Evolution. 23 : 261267. Binkley, C.S. 1997. Preserving nature through intensive plantation forestry: the case of forestland allocation with illustrations from British Columbia. Forestry Chronicle 73: 553559. Boltz, F., Carter, D.R., Holmes, T.P., Pereira Jr., R. 2001. Financial returns under uncertainty for conventional and reducedimpact logging in permanent production forests of the Brazilian Amazon. Ecological Economics 39: 387398. Borchert, R. 1994. Water storage in soil or tree stems determines phenology and distribution of tropical dry forest trees. Ecology 75: 14371449.

PAGE 122

122 Borchert, R., Meyer, S.A., Fleger, R.S., Porter Bolland, L. 2004. Environmental control of flowering periodicity in Costa Rican and Mexican tropical dry fores ts. Global Ecology and Biogeography 13: 409425. Boscolo, M., Buongiorno, J. 1997. Managing a tropical rainforest for timber, carbon storage, and tree diversity. Commonwealth Forestry Review 76 : 246254. Boscolo, M., Vincent, J.R. 2003. Nonconvexities in t he production of timber, biodiversity, and carbon sequestration. Journal of Environmental Economics and Management 46: 251268. Bray, D.B., Merino Perz, L., Negreros Castillo, P., SeguraWarnholtz, G., Torres Rojo, J.M. Vester, H.F.M. 2003. Mexicos comm unity managed forests as a global model for sustainable landscapes. Conservation Biology 17: 672 673. Brearley, F.Q., Proctor, J., Suriantata, L., Nagy, G., Dalrymple, Voysey, B.C. 2007. Reproductive phenology over a 10year period in a lowland evergreen rain forest of central Borneo. Journal of Ecology 95: 828839. Brechin, S.R., Wilshusen, P.R. Fortwangler, C.L. West P.C Beyond the square wheel: Toward a more comprehensive understanding of biodiversity conservation as a social and political process. Society and Natural Resources 15 : 41 64. Brooks, T.M., Wright, S.J., Sheil, D. 2009. Evaluating the success of conservation actions in safeguarding tropical forest biodiversity. Conservation Biology 23: 14481457. Bruna, E.M., 2003. Are plant populations in fragmented habitats recruitment limited? Tests with an Amazonian herb. Ecology. 84, 932947. Bruna, E.M., Kress, W.J. 2002. Habitat fragmentation and the demographic structure of an Amazonian understory herb ( Heliconia acuminata).Conservation Biology 1 6 : 12561266. Buongiorno, J. and Gilless, J.K. 2000. Decision Methods for Forest Resource Management. Academic Press, Elsevier Science, San Diego, CA. 439p. Buongiorno, J.L, Peyron, F., Houllier, Bruciamacchie, M. 1995. Growth and management of mixedspeci es, unevenaged forests in the French Jura: implications for economic returns and tree diversity. Forest Science 41: 397 429. Campos, J.J., Finegan, B. Villalobos R 2001. Management of Goods and Services from Neotropical Forest Biodiversity: Diversified Forest Management in Mesoamerica. CBD Technical Series No. 3, Secretariat of the Convention on Biological Diversity (SCBD), Montreal, pp. 5 16. Carslake, D., S. Townley, D.J. Hodgson. 2009. Patterns and rules for sensitivity and elasticity in population projection matrices. Ecology 90: 32583267.

PAGE 123

123 Castro, E.R., Galetti, M., Morellato, L.P.C., 2007. Reproductive phenology of Euterpe edulis (Arecaceae) along a gradient in the Atlantic rainforest of Brazil. Australian Journal of Botany. 55, 725735. Caswell, H. 2001. Matrix Population Models: Construction, Analysis and Interpretation. Second Edition. Sinauer Associates, Inc. Publishers. Sunderland, MA. 722p. Caswell, H. 2010. Life table response experiment analysis of the stochastic growth rate. Journal of Ecol ogy 98: 324333. Chopra, K. 1993. The value of nontimber forest products: an estimation for tropical deciduous forests in India. Economic Botany 47: 251257. Clark, D.A., Clark, D.B., 1990. Distribution and effects on tree growth of lianas and woody hemiepiphytes in a Costa Rican tropical wet forest. Journal of Tropical Ecology. 6, 321331. Clay, J.W., Clement, C.R. Sampaio P.B 2000. Biodiversidade Amaznica: exemplos e estratgias de utilizao. Manaus: Programa de Desenvolvimento Empresarial e Tecnolgico. 409 p. Cleland, E.E., Chuine, I., Menzel, A., Mooney, H.A., Schwartz, M.D. 2007. Shifting plant phenology in response to global change. Trends in Ecology and Evolution. 22: 357365. Cochrane, M.A., Alencar, A. Schulze, M.D., Souza, C.M., Nepstad, D.C., Lefebvre, P., Davidson, E.A. 1999. Positive feedbacks in the fire dynamics of closed canopy tropical forests. Science 284 : 18321835. Cox, P.M., Harris, P.P., Huntingford, C., Betts, R.A., Collins, M., Jones, C.D., Jupp, T.E., Marengo, J.A., Nobre, C.A. 2008. Increasing risk of Amazonian drought due to decreasing aerosol pollution. Nature 453: 212215. Crouse, D., Crowder, L., and Caswell, H. 1987. A stagebased population model for loggerhead sea turtles and implications for conservation. Ecology 6 8 : 14121423. Crowder, L.B., Crouse, D.T., Heppell, S.S., Martin, T.H. 1994. Predicting the impact of turtle excluder devices on loggerhead sea turtle populations. Ecological applications 4: 437445. Daily,G.C. 1997. Natures services: societal dependence on natural ecosystems. Washington, D.C., Island Press. Davidson, R., Jacquemyn, H., Adriaens, D., Honnay, O., de Kroon, H., Tuljapurkar, S. 2010. Demographic effects fo extreme weather events on a short lived calcareous grassland species: stochastic life t able response experiments. Journal of Ecology 98: 255267.

PAGE 124

124 Davies, S.J., Palmiotto, P.A., Ashton, P.S., Lee, H.S., LaFrankie, J.V., 1998. Comparative ecology of 11 sympatric species of Macaranga in Borneo: tree distribution in relation to horizontal and vertical resource heterogeneity. Journal of Ecology. 86, 662673. da Silva Dias, A., Campos, J.J., Villalobos, R., Louman, B., Gonalvez, L. 2002. Manejo florestal diversificado en una comunidad riberea de la Amazona bresilea: consideraciones socials y s ilviculturales. Revista Forestal Centroamericana 38: 7884. Dayanandan, S., Dole, J., Bawa, K., Kesseli, R. 1999. Population structure delineated with microsatellite markers in fragmented populations of a tropical tree, Carapa guianensis (Meliaceae). Mole cular Ecology. 8: 15851592. De Graaf, N.R., Poels, R.L.H., Van Rampaey, R.S.A.R., 1999. Effect of silvicultural treatment on growth and mortality of rainforest in Surinam over long periods. Forest Ecology and Management 124, 123135. Dekker, M., De Graaf N.R., 2003. Pioneer and climax tree regeneration following selective logging with silviculture in Suriname. Forest Ecology and Management 172, 183190. De Kroon, H., Plaisier, A., van Groenendael, J., Caswell, H. 1986. Elasticity: the relative contribut ion of demographic parameters to population growth rate. Ecology 67:14271431. DeSteven, D., Windsor, D.M., Putz, F.E., de Leon, B. 1987. Vegetative and reproductive phonologies of a palm assemblage in Panama. Biotropica 19: 342356. Daz, M., Van der Pilj L. 2005. The significance of episodic rains for reproductive phenology and productivity of trees in semiarid regions of northwestern Venezuela. Trees Structure and Function 19: 336348. Doak, D.F., Morris, W.F., Pfister, C., Kendall, B.E., Bruna, E.M. 20 05. Correctly estimating how environmental stochasticity influences fitness and population growth. The American Naturalist 166: E14E21. Dykstra, D.P., Heinrich R 1996. FAO Model Code of Forest Harvesting Practice. Food and Agricultural Organization of t he United Nations, Rome, Italy. Elzinga, J.A., Atlan, A., Biere, A., Gigord, L., Weis, A.E., Bernasconi, G. 2007. Time after time: flowering phenology and biotic interactions. Trends in Ecology and Evolution 22: 432439. Emanuel, P.C., Shackleton, C.M., Ba xter, J.S. 2005. Modeling the sustainable harvest of Sclerocarya birrea subsp. Caffra fruits in the South African lowveld. Forest Ecology and Mangement 214: 91103.

PAGE 125

125 Engel, S., Pagiola, S., Wunder S.. 2008. Designing payments for environmental services in theory and practice: an overview of the issues. Ecological Economics 65: 663674. Fearnside, P.M. 2005. Deforestation in Brazilian Amazonia: History, rates, and consequences. Conservation Biology. 19(3):680688. Fenner, M. 1998. The phenology of growth and reproduction in plants. Perspectives in Plant Ecology, Evolution and Systematics 1: 7891. Ferraz, I.D.K., Camargo, J.L.C., Sampaio, P.T.B., 2002. Sementes e plntulas de andiroba ( Carapa guianensis Aubl. E C. procera D.C. Meliaceae) Aspectos botnicos ecolgicos e tecnolgicos. Acta Amazonica. 32, 661647. Forbes, V.E., Olsen, M., Palmqvist, A., Calow, P. 2010. Environmentally sensitive lifecycle traits have low elasticity: implications for theory and practice. Ecological Applications 20: 14491455. Forget, P.M., 1996. Removal of seeds of Carapa procera (Meliaceae) by rodents and their fate in Rainforest in French Guiana. Journal of Tropical Ecology. 12, 751761. Fournier, L.A., 2003. Species Description Carapa guianensis In: Vozzo, J.A. Tropical Tre e Seed Manual. USDA Forest Service, Washington DC. Agricultural Handbook Number 721. Fox, J.C., Hamilton, F., Ades P.K 2008. Models of tree level hollow incidence in Victorian State forests. Forest Ecology and Management 255: 28462857. Franco, M., Silvertown, J., 2004. A comparative demography of plants based upon elasticities of vital rates. Ecology 85: 531538. Gaillard, J M., Festa Bianchet, M., Yoccoz, N.G. 1998. Population dynamics of large herbivores: variable recruitment with constant adult surviv al. Trends in Ecology and Evolution 13: 58 63. Gaoue, O.G., Ticktin, T. 2009. Fulani knowledge of the ecological impacts of Khaya senegalensis (Meliaceae) foliage harvest in Benin and its implications for sustainable harvest. Economic Botany 63(3): 256-270. Garca Fernndez, C., Ruiz Prez, M., Wunder S 2008. Is multiple use forest management widely implementable in the tropics? Forest Ecology and Management 256 : 1468 1476. Gardingen, P.R., van, Valle, D., Thompson, I. 2006. Evaluation of yield regulation options for primary forest in Tapajs National Forest, Brazil. Forest Ecology and Management 231 : 184 195.

PAGE 126

126 Gardner, T.A., Barlow, J., Chazdon, R. Ewers, R. Harvey, C.A., Peres, C.A., Sodhi, N.S. 2009. Prospects for tropical biodiversity conservation in a humanmodified world. Ecology Letters 12: 561582. Gartlan, J.S., Newbery, D.McC., Thomas, D.W., Waterman, P.G., 1986. The influence of topography and soil phosphorus on the vegetation of Korup Forest Reserve, Cameroun. Vegetatio. 65, 131148. Grauel, W .T., Putz, F.E., 2004. Effects of lianas on growth and regeneration of Prioria copaifera in Darien, Panama. Forest Ecology and Management. 190, 99108. Greene, D.F., Johnson, E.A., 1994. Estimating the mean annual seed production of trees. Ecology. 75, 642 647. Gross, K., Morris, W.F., Wolosin, M.S., Doak D.F 2006. Modeling vital rates improves estimation of population projection matrices. Population Ecology 48: 79 89. Guariguata, M.R., Cronkleton, P., Shanley, P., Taylor, P.L. 2008. The compatibility of timber and nontimber forest product extraction and management. Forest Ecology and Management, 256: 14771481. Guariguata, M.R., Claire, H.A L., Jones, G., 2002. Tree seed fate in a logged and fragmented forest landscape, Northeastern Costa Rica. Biotropic a. 34, 405415. Guariguata, M.R., GarcaFernndez, C., Sheil, D., Nasi, R., Herrero Juregui, C., Cronkleton, P. Ingram V 2010. Compatibility of timber and nontimber forest product management in natural tropical forest: Perspectives, challenges, and o pportunities. Forest Ecology and Management 259 : 237 245. Guedes, M.C., Souto, .B., Correa, C., Gomes, H.S.R., 2008. Produo de sementes e oleo de andiroba ( Carapa guianensis Aubl.) em area de vrzea do Amap. In: Wadt, L.H.O., Anais do 1 de Produtos Florestais Nomadeireiros na Amaznia, Rio Branco, Acre. Gunatilake, H.M., Senaratne, D.M.A.H., Abeygunawardena, P. 1993. Role of nontimber forest products in the economy of peripheral communities of Knuckles National Wilderness area of Sri Lanka: A farming systems approach. Economic Botany 47: 275281. Hall, P., Orrell, L.C., Bawa, K.S., 1994. Genetic diversity and mating system in a tropical tree, Carapa guianensis (Meliaceae). American Journal of Botany. 81, 11041111. Harper, J.L., 1977. Population Biology of Plants, Academic Press, New York. Harvey, C., Komar, O., Chazdon, R., Ferguson, B., Finegan, B., Griffith, D., Martinez Ramos, M., Morales, H., Nigh, R., Soto Pinto L 2008. Integrating agricultural landscapes with biodiversity conservation in the Mesoamerican hotspot. Conservation Biology 22: 8 15.

PAGE 127

127 Herrera, C.M., Jordano, P., Guitin, J., Traveset, A., 1998. Annual variability in seed production by woody plants and the masting concept: reassessment of principles and relationship to pollination and seed dispersal. The American Naturalist. 152, 576594. Herrero Juregui, C., GarcaFernndez, C., Sist, P., Casado, M. 2009. Conflict of use for multi purpose tree species in the state of Par, eastern Amazonia, Brazil. Biodiversity and Conservation 18 : 10191044. Hett, J.M., 1971. A dynamic analysis of age in sugar maple seedlings. Ecology. 52, 10711074. Heywood, V.H., Iriondo, J.M 2003. Plant conservation: old problems, new perspectives. Biological Conservation 113 : 321335. Holmes, T.P., Blate, G.M., Zweede, J.C., Perreira Jr., R., Barreto, P., Boltz, F., Bauch, R. 2002. Financial and ecological indicators of reduced impact logging performance in the eastern Amazon. Forest Ecology and Management 163: 93110. Horvitz, C. C., Schemske D. W 1995. Spatiotemporal variation in demographic transitions of a tropical under story herb: projection matrix analysis. Ecological Monographs 65:155192. Hubbell, S.P., Foster, R.B., 1986. Commonness and rarity in a neotropical forest: implications for tropical tree conservation. In: Soule, M. (Ed.), Conservation Biology: the Science of Scarcity and Diversity. Sinauer Associates, Sunderland, Massachusetts, pp. 205231. Huerte, A.R., Didan, K., Shimabukuro, Y.E., Ratana, P., Saleska, S.R., Hutyra, L.R., Yang, W., Nemani, R.R., Myneni, R. 2006. Amazon rainforests greenup with sunlight in dry season. Geophysical Research Letters 33, L06405, doi: 10.1029/2005GL025583. Instituto Nacional de Meteorologia (INMET) 2008. Federal University of Acre, Brazil. http://www.inmet.gov.br Intergovernmental Panel on Climate Change (IPCC). 2007. Climate change 2007: the physical science basis Contribution of work ing group I to the fourth assessment report of the IPCC. Cambridge, UK. Cambridge University Press. Jacobson, M. 1998. Comparing values of timber production to agricultural crop production. University of Florida Cooperative Extension Service Institute of F ood and Agricultural Sciences bulletin. Janzen, D.H., 1971. Euglossine bees as long distance pollinators of tropical plants. Science. 171, 203205.

PAGE 128

128 Janzen, D.H. 1986. The eternal external threat. Pages 237257 in M.E. Soul, editor. Conservation Biology. Sinauer Associates, Sunderland, Massachusetts. Kainer, K.A., Wadt, L.H.O., Gomes Silva, D.A.P., Capanu, M., 2006. Liana loads and their association with Bertholletia excelsa fruit and nut production, diameter growth and crown attributes. Journal of Tropic al Ecology. 22, 147154. Kainer, K.A., Wadt, L.H.O., Staudhammer, C.L., 2007. Explaining variation in Brazil nut fruit production. Forest Ecology and Management. 250, 244255. Kawecki, T.J. 2008. Adaptation to marginal habitats. Annual Review of Ecology, E volution and Systematics 39: 321342. Kelly, D., 1994. The evolutionary ecology of mast seeding. Trends in Ecology and Evolution. 9, 465470. Kelly, D., Harrison, A.L., Lee, W.G., Payton, I.J., Wilson, P.R., Schauber, E.M., 2000. Predator satiation and ext reme mast seeding in 11 species of Chionochloa (Poaceae). Oikos. 90, 477488. Kelly, D., Sork, V.L., 2002. Mast seeding in perennial plants: Why, How, Where? Annual Reviews in Ecology and Systematics. 33, 427 447. Kelly, D., Turnbull, M.H., Pharis, R.P., Sarfati, M.S., 2008. Mast seeding, predator satiation, and temperature cues in Chionochloa (Poaceae). Population Ecology 50, 343355. Klemperer, D.W. 1996. Economics of forestland use and Evenaged rotations. Forest Resource Economics and Finance, McGraw Hill, Inc. NY 551p. Klimas, C.A., Kainer, K.A., Wadt, L.H.O., 2007. Population structure of Carapa guianensis in two forest types in the southwestern Brazilian Amazon. Forest Ecology and Management 250, 256265. Koenig, W.D., Knops J.M.H 2000. Patterns of annual seed production by northern hemisphere trees: a global perspective. American Naturalist 155 : 59 69. Koren, I., Remer, L.A., Longo, K. 2007. Reversal fof trend of biomass burning in the Amazon. Geophysical Research Letters 34 : L20404, doi: 10.1029/2007GL031530. Kozlowski, T.T., Pallardy, S.G., 2002. Acclimation and adaptive responses of woody plants to environmental stresses. Botanical Review. 68, 270334. Lamb, D., Erskine, P.D., Parrotta J.A 2005. Restoration of degraded tropical forest landsca pes. Science 310 : 16281632. Laurance, W.F., 1999. Reflections on the tropical deforestation crisis. Biological Conservation. 91, 109117.

PAGE 129

129 Leftkovitch, L.P. 1967. A theoretical evaluation of population growth after removing individuals from some age groups Bulletin of Entomological Research 57:437445. Leite, A.M.C., 1997. Ecologia de Carapa guianensis Aublet (Meliaceae) Andiroba. Tese (Doutorado em Botnica). Instituto Nacional de Pesquisas da Amaznia. Manaus, 180p. Lentini, M., Verssimo, A., Sobral, L. 2003. Fators Florestais da Amaznia. IMAZON. Belm, Brazil. 110p. Lieberman, M., Lieberman, D., Peralta, R., Hartshorn, G.S., 1995. Canopy closure and the distribution of tropical forest trees at La Selva, Costa Rica. Journal of Tropical Ecology. 11, 1 61178. Lima, E., Merry, F. Nepstad, D., Amacher, G., AzevedoRamos, C., Lefebvre, P., Resque, F 2006. Searching for sustainability. Environment 48 : 26 38. Lindenmayer, D.B., Cunningham, R.B., Donnelly, C.F., Tanton, M.T., Nix H.A 1993. The abundance and development of cavities in eucalyptus trees: a case study in the montane forests of Victoria, southeastern Australia. Forest Ecology and Management 60: 77104. Lindenmayer, D.B., Cunningham, R.B., Pope, M.L., Gibbons, P., Donnelly C.F 2000. Cavity si zes and types in Australian eucalypts from wet and dry forest types a simple rule of thumb for estimating size and number of cavities. Forest Ecology and Management 137: 139150. Lipper, L. 2000. Forest degradation and food security. Unasylva 202: onlin e URL: http://www.fao.org/docrep/X7273E/x7273e05.htm Londres, M., 2004. Diagnstico Participativo do Sistema Produtivo da Andiroba ( Carapa sp.) e Estudo Botnico da Espcie, nas Comunidades So Joodo Jaburu e Fortaleza do Jaburu, Municpio de Gurup, PA. Universidade de So Paulo, Piracicaba. Londres, M., 2009. Population structure and seed production of Carapa guianensis in three floodplain forest types of the Amazon estuary. M.S. Thesis, University Florida, Gainesville, FL. 56 p. Mabberley, D.J. 1993. The Plant Book A Portable Dictionary of the Higher Plants. Department of Plant Sciences, University of Oxford, Cambridge University Press, Great Britain. Macpherson, A.J. 2007. Following the rules: A bioeconomic policy simulation of a Brazilian forest concession. Dissertation, University of Florida, Gainesville 192p.

PAGE 130

130 Marengo, J.A., Nobre, C.A., Tomasella, J., Oyama, M.D., Sampaio de Oliveira, G., de Oliveira, R., Camargo, H., Alves, L.M., Brown, I.F. 2008. The drought of Amazonia in 2005. Journal of Climat e 21: 495516. Marshall, E., Newton, A.C., Schreckenberg, K. 2003. Commercialization of nontimber forest products: first steps in anlalysing the factors influencing success. International Forestry Review 5: 128137. Martini, A.M.Z., Rosa, N.A., Uhl, C. 19 94. An attempt to predict which Amazonian tree species may be threatened by logging activities. Environmental Conservation 21: 152162. MATLAB. 2002. MATLAB. Version 6.5. The MathWorks, Natick, Massachusetts. Maus M.M., 2006. Estratgias reprodutivas de espcies arbreas e a sua importncia para o manejo e conservao florestal: Floresta Nacional do Tapajs (BelterraPA). Tese (Doutorado) Universidade de Braslia. Instituto de Cincias Biolgicas. Braslia, DF. 206 p. McHargue, L.A., Hartshorn, G.S., 19 83a. Seed and seedling ecology of Carapa guianensis Turrialba. 33, 399404. McHargue, L.A., Hartshorn, G.S., 1983b. Carapa guianensis In: Janzen, D.H. (Ed.), Costa Rican natural history, University of Chicago Press, Chicago, IL. pp. 206207. McLaren, K.P ., McDonald, M.A. 2005. Seasonal patterns of flowering and fruiting in a dry tropical forest in Jamaica. Biotropica 37: 584590. Medida Provisria N 67, August 24, 2001 Meekins, J.F., McCarthy, B.C., 2002. Effect of population density on the demography of an invasive plant ( Alliaria petiolata Brassicaceae) population in a southeastern Ohio forest. American Midland Naturalist 147: 256278. Meijaard, E. D., Sheil, D., Nasi, R., Augeri, D., Rosenbaum, B., Iskandar, D., Setyawati, T., Lammertink, M.J., Rachmatika, I., Wong, A., Soehartono, T., Stanley, S., OBrien, T. 2005. Life after logging: reconciling wildlife conservation and production forestry in Indonesian Borneo. CIFOR, Bogor, Indonesia. http://www.cifor.cgiar.org/publications/pdf_files/books/B Meijaard0501E0.pdf Menton, M.C.S., Merry, F.D., Lawrence, A., Brown N 2009. Company Community Logging Contracts in Amazonian Settlements: Impacts on Livelihoods and NTFP Harvests. Ecology and Society 14: 39. Merry, F., Soares Filho, B., Nepstad, D., Amac her, G., Rodrigues H 2009. Balancing conservation and economic sustainability: the future of the Amazon timber industry. Environmental Management 44: 395 407.

PAGE 131

131 Mollinedo, A. 2000. Beneficios Sociales y Rentabilidad del Manejo Forestal Comunitario en dos reas de la Reserva de la Biosfera Maya, Petn, Guatemala. M.Sc. CATIE, Turrialba, Costa Rica. Mollinedo, A., Campos, J.J., Kanninen, M., Gmez M 2001. Benficios socials y econmicos del bosque en la Reserva de la Biosfera Maya, Petn, Guatemala. Revista Forestal Centroamericana. CATIE 34 : 57 60. Monks, A., Kelly, D., 2006. Testing the resourcematching hypothesis in the mast seeding tree Nothofagus truncate (Fagaceae). Austral Ecology. 31, 366375. Morris, W. F., Doak, D.F. 2002. Quantitative Conservation Biology: Theory and Practice of Population Viability Analysis. Sinauer Associates Inc. Publishers. Sunderland, MA. 480p. Morris, W.F., Doak, D.F., 2005. How general are the determinants of the stochastic population growth rate across nearby sites? Ecolog ical Monographs 75: 119137. Morsello, C. 2006. Company community non timber forest product deals in the Brazilian Amazon: A review of opportunities and problems. Forest Policy and Economics 8 : 485494. Morton, D.C., DeFries, R.S., Shimabukuro, Y.E., Ander son, L.O., Arai, E., del Bon Espirito Santo, F., Freitas, R., Morisette J 2006. Cropland expansion changes deforestation dynamics in the southern Brazilian Amazon. Proceedings of the National Academy of Science USA 103 : 14637 14641. NabeNielsen, J. Kollmann, J., PeaClaros, M., 2009. Effects of liana load, tree diameter and distances between conspecifics on seed production in tropical timber trees. Forest Ecology and Management. 257, 987993. Nantel, P., Gagnon, G., and Nault, A. 1996. Population viabil ity analysis of American Ginseng and Wild Leek harvested in stochastic environments. Conservation Biology 10: 608 621. Nepstad, D., Carvalho, G., Barros, A.C., Alencar, A., Capobianco, J.P., Bishop, J., Moutinho, P. Lefebvre, P., Lopes Silva, Jr., U. P rins E 2001. Road paving, fire regime feedbacks, and the future of Amazon forests. Forest Ecology and Management 154:395407. Nepstad, D., Soares Filho, B.S., Merry, F., Lima, A., Moutinho, P., Carter, J., Bowman, M., Cattaneo, A., Rodrigues, H., Schwartzman, S., McGrath, D.G., Stickler, C.M., Lubowski, R., Piris Cabezas, P., Rivero, S., Alencar, A., Almeida, O., Stella, O. 2009. The end of deforestation in the Brazilian Amazon. Science 326 : 13501351.

PAGE 132

132 Nepstad, D.C., Verissimo, A., Alencar, A., Nobre, C., Lima, E., Lefebvre, P., Schlesinger, P., Potter C., Moutinho P., Mendoza, E., Cochrane, M., and Brooks, V. 1999. Largescale impoverishment of Amazonian forest by logging and fire. Nature. 398 : 505 508. Newbery, D.McC., Gartlan, J.S., McKey, D.B., Wate rman, P.G., 1986. The influence of drainage and soil phosphorus on the vegetation of DoualaEdea Forest Reserve, Cameroun. Vegetatio. 65, 149162. Norton, D.A., Kelly D 1988. Mast seeding over 33 years by Dacrydium cupressinum Lamb. (rimu) (Podocarpaceae) in New Zealand: the importance of economies of scale. Functional Ecology 2 : 399 408. OBrien, J.J., Oberbauer, S.F., Clark, D.B., Clark, D.A., 2008. Phenology and stem diameter increment seasonality in a Costa Rican wet tropical forest. Biotropica. 40, 1 51159. Oliveira, M..N.d. 2000. Sustainable forest managers for small farmers in Acre state in the Brazilian Amazon. Tropical Forest Update 8: 57. Oliveira, M..N.d., Braz, E.M., Burslem, D.F.R.P., Swaine, M.D. 1998. Small scale natural forest management : a new model for small farmers in the Brazilian Amazon. Tropical Forest Update 8: 57 Olmsted, I., Alvarez Buylla, E.R.. 1995. Sustainable harvesting of tropical trees: Demography and matrix models of two palm species in Mexico. Ecological Applications. 5(2):484 500. Padoch, C. 1988. Aguaje ( Mauritia flexuosa L.f.) in the economy of Iquitos, Peru. Advances in Economic Botany 6: 214224. Padoch, C., PinedoVasquez, M. 1996. Smallholder forest management: looking beyond nontimber forest products. In: Ruiz Prez M and Arnold JEM (eds) Current Issues in Non Timber Forest Products Research. CIFOR, Bogor, Indonesia. Panayotou, T., Aston, P.S. 1993. Not by timber alone: economics and ecology for sustaining tropical forests. Island Press, Washington. Pascarella, J. B., Horvitz, C.C., 1998. Hurricane disturbance and the population dynamics of a tropical understory shrub: megamatrix elasticity analysis. Ecology 79: 547563. Pattanayak, S.K., Sills E.O. 2001. Do tropical forests provide natural insurance? The microecon omics of non timber forest product collection in the Brazilian Amazon. Land Economics 77: 595 613. Pearce, D. 1996. Can nonmarket values save the worlds forests? In: International Symposium on the non market Benefits of Forestry, Edinburg.

PAGE 133

133 Pearce, D., P utz, F.E., Vanclay J.K 2003. Sustainable forestry in the tropics: panacea or folly? Forest Ecology and Management 172 : 229 247. Pennington, T.D., 1981. Meliaceae (Flora Neotropica, Monograph No. 28). New York Botanical Garden Pr. Department, New York, 47 0 p. Peres, C.A., Baider, C., Zuidema, P.A., Wadt, L.H.O., Kainer, K.A., Gomes Silva, D.A.P., Salomo, R.P., Simes, L.L., Francisiosi, E.R.N., Valverde, F.C., Gribel, R. Shepard, Jr., G.H., Kanashiro, M., Coventry, P., Yu, D.W., Watkinson, A.R., Frecklet on, R.P., 2003. Demographic threats to the sustainability of Brazil nut exploitation. Science 302, 21122114. Peters, C.M., 1996. Chapter I III: Introduction to the ecology of tropical forest resources. In: The ecology and management of nontimber forest resources. World Bank Technical Paper No. 322. The World Bank, Washington D.C., pp. 597. Phillips, O.L., Aragao, L.E.O.C., Lewis, S.L., Fisher, J.B. et al. 2009. Drought sensitivity of the Amazon Rainforest. Science 323: 1344 1347. Piovesan G., Adams, J. M., 2001. Masting behaviour in beech: linking reproduction and climatic variation. Canadian Journal of Botany. 79, 10391047. Pfister, C.A. 1998. Patterns of variance in stagestructured populations: evolutionary predictions and ecological implications. Pr oceedings of the National Academy of Sciences USA 95 : 213 218. Pinard, M.A., F.E. Putz, J. Tay, T.E. Sullivan. 1995. Creating timber harvesting guidelines for a reducedimpact logging project in Malaysia. Journal of Forestry 93: 4145. Pinero, D., Martinez Ramos, M. Sarukhan, J 1984. A population model of Astrocarvum mexicanum and a sensitivity analysis of its finite rate of increase. Journal of Ecology 72:977 992. Pinto, L.F.G., Shanley, P., Gomes, A.P.C., Robinson, D. 2008. Experience with NTFP certifi cation in Brazil. Forests, Trees and Livelihoods 18: 3754. Plowden, C., 2004. The ecology and harvest of andiroba seeds for oil production in the Brazilian Amazon. Conservation & Society, 2, 251272. Poore, M.E.D., 1968. Studies in Malaysia rain forest. I The forest on Triassic sediments in Jengka Forest Reserve. Journal of Ecology. 56, 143196. Putz, F.E., Dykstra, D.P., Heinrich, R. 2000. Why poor logging practices persist in the tropics. Conservation Biology 14: 951956. R Development Core Team, 2006. R: A Language and Environment for Statistical Computing. Vienna, Austria, ISBN: 3 900051 070, http://www.R project.org

PAGE 134

134 Ramula, S., Rees, M., Buckley, Y.M. 2009. Integral projection models perform better for small demographic data sets than matrix populat ion models: a case study of two perennial herbs. Journal of Applied Ecology 46: 10481053. Ranklin, J.M., 1978. The influence of seed predation and plant competition on three species abundances in two tropical rain forest communities in Trinidad, West Indi es. Ann Arbor: University of Michigan. Ratsirarson, J., Silander, J.A., and Richard, A.F. 1996. Conservation and management of a threatened Madagascar palm species, Neodysis decaryi Jumelle. Conservation Biology 10: 40 52. Rees, M., Kelly, D., Bjrnstad, O.N., 2002. Snow Tussocks, Chaos, and the Evolution of Mast Seeding. The American Naturalist. 160, 4459. Resoluo N. 406. De 02 de Fevereiro de 2009, Ministrio do Meio Ambiente, Conselho Nacional do Meio Ambiente. Brazil Rivera, G., Borchert, R. 2001. I nduction of flowering in tropical trees by a 30min reduction in photoperiod: evidence from field observations and herbarium specimens. Tree Physiology 21: 201212. Rivera, G., Elliot, S., Caldas, L.S., Nicolossi, G., Coradin, V.T.R., Borchert, R. 2002. In creasing day length induces spring flushing of tropical dry forests in the absence of rain. Trees Structure and Function 16: 445 456. Rockwell, C., Kainer, K.A., Marcondes, N., Baraloto, C. 2007. Ecological limitations of reducedimpact logging at the smal lholder scale. Forest Ecology and Management 238: 365374. Rodrigues, R.M. 1989. A Flora da Amaznia. Belm: CEJUPE. Utilidades industrias. Plantas Medicinais. Rodgrigues, A.S.L., Ewers, R.M., Parry, L., Souza Jr., C., Verssimo, A., Balmford, A., 2009. Boom and Bust Development Patterns Across the Amazon Deforestation Frontier. Science. 324, 14351437. Rogers Bennett, L., Leaf R.T. 2006. Elasticity analyses of sizebased red and white abalone matrix models: management and conservation. Ecological Applic ations 16:213 224. Ribeiro do Valle, D., Staudhammer, C.L., Cropper, Jr. W.P. 2007. Simulating nontimber forest product management in tropical mixed forests. Journal of Forestry 105:301 306

PAGE 135

135 Ros Tonen, M.A.F., Andel, T.v., Morsello, C., Otsuki, K., Rosendo S., Scholz I. 2008. Forest related partnerships in Brazilian Amazonia: There is more to sustainable forest management than reduced impact logging. Forest Ecology and Management 256 : 1482 1497. Saleska, S.R., Didan, K., Huerte, A.R., da Rocha, H.R. 2007. Amazon forests greenup during 2005 drought. Science: 318: 612. Salick, J., Mejia, A., Anderson, T. 1995. Nontimber forest products integrated with natural forest management. Rio San Juan, Nicaragua. Ecological Applications 5(4): 878895. doi:10.2307/226 9339. Samanta, A., Ganguly, S., Hashimoto, H., Devadiga, S., Vermonte, E., Knyazikhin, Y., Nemani, R.R., Myneni, R.B. 2010. Amazon forests did not greenup during the 2005 drought. Geophysical Research Letters 37, L05401, doi:10.1029/2009GL042154 Scales, B .R., Marsden, S.J. 2008. Biodiversity in small scale tropical agroforests: a review of species richness and abundance shifts and the factors influencing them. Environmental Conservation 35: 160 172. Schauber, E.M., Kelly, D., Turchin, P., Simon, C., Lee, W .G., Allen, R.B., Payton, I.J., Wilson, P.R., Cowan, P.E., Brockie, R.E., 2002.Masting by eighteen New Zealand plant species: the role of temperature as a synchronizing cue. Ecology. 83, 12141225. Scherr, S., White, A., Kaimowitz, D. 2003. Making markets work for communities. International Forestry Review 5(1): 6773. Schmitt, C.B., Belokurov, A., Besanon, C., Boisrobert, L., Burgess, N.D., Campbell, A., Coad, L. Fish, L., Gliddon, D., Humphries, K. Kapos, V., Loucks, C., Lysenko, I., Miles, L., Mills, C., Minnemeyer, S., Pistorius, T., Ravilious, C., Steininger, M., Winkel G. 2008. Global ecological forest classification and forest protected area gap analysis. Analysis and recommendations in view of the 10% target for forest protection under the Convention on Biological Diversity (CBD). In University of Friedberg, Freidberg. Segura, M., Kanninen, M. 2005. Allometric models for tree volume and total aboveground biomass in a tropical humid forest in Costa Rica. Biotropica 37 : 2 8. Shanley, P. Luz, L., Sw ingland, I.A. 2002. The faint promise of a distant market: a survey of Belms trade in nontimber forest products. Biodiversity and Conservation 11: 615636. Shanley, P., Medina, G., Cymerys, M., Galvo, J., 1998. Frutferas da Mata na Vida Amaznica. Bel m: CIFOR, Imazon, 127 p.

PAGE 136

136 Shandley, P., Medina, G., 2005. Fruitferas e plantas teis na vida Amaznica. Belm: CIFOR, Imazon. 304 p. Sheil, D. 1995. A critique of permanent plot methods and analysis with examples from Budongo Forest, Uganda. Forest Ecology and Management. 77, 11 34. Sheil, D. 1997. Long term growth and rainfall in a Ugandan moist forest: seasonal rhythms and flexing stems. Commonwealth Forestry Review. 76, 121127. Silva, A.C.C., 2009. Remoo e destino de sementes de Carapa guianensis A ubl. (Meliaceae) e Bertholletia excelsa Bonpl. (Lecythidaceae) no sudoeste do Estado do Acre, Brasil. M.S. Thesis (Ecology and Natural Resource Management) Federal University of Acre (UFAC). 153 p. Silva, J.N.M., Carvalho, J.O.P.de, Lopes, J.doC.A., Carval ho, M.S.P.de. 1984. Equaes de volume para a Floresta Nacional do Tapajs (Volume equations for the Tapajs National Forest). Boletim de Pesquisa Florestal, Columbo 8/9 : 50 63. Silva, J.N.M., De Carvalho, J.O.P., Lopes, J.D.A., De Almeida, B.F., Costa, D.H.M., De Oliveira, L.C., Vanclay, J.K., Skovsgaard, J.P., 1995. Growth and yield of a tropical rainforest in the Brazilian Amazon 13 years after logging. Forest Ecology and Management 71, 267274. Smith, D.M., Larson B.C., Kelty, M.J., Ashton, P.M.S., 199 7. The Practice of Silviculture: Applied Forest Ecology, ninth ed. John Wiley & Sons, Inc., New York, NY. Snook, L.K. 2000. Utilization and management of timber and nontimber forest resources in the forest Ejidos of Quintana Roo Mexico. Case studies on combined Timber Non timber management. 615, FAO, Rome, Italy. Snook, L.K., CmaraCabrales, L., Kelty, M.J., 2005. Six years of fruit production by mahogany trees ( Swietenia macrophylla King): patterns of variation and implications for sustainability. Fores t Ecology and Management. 206, 221235. Soehartono, T., Newton, A.C. 2001. Reproductive ecology of Aquilaria spp. In Indonesia. Forest Ecology and Management 152: 5971. Sork, V.L., 1993. Evolutionary ecology of mast seeding in temperate and tropical oaks ( Quercus spp.). Vegetatio. 107/108, 133147. Sork, V.L., Bramble, J.E., 1993. Prediction of acorn crops in three species of North American oaks: Quercus alba, Q. rubra and Q. velutina Annales des Sciences Forestieres. 50(Suppl.), 128136. Stevens,G.C., 1 987. Lianas as structural parasites: the Bursera simaruba example. Ecology. 68, 7781.

PAGE 137

137 Stevenson, P.R., 2004. Phenological patterns of woody vegetation at Tinigua Park Columbia: Methodological comparisons with emphasis on fruit production. Caldasia. 26, 12 5 150. Stevenson, P.R., Castellanos, M.C., Corts, A.I., Link, A., 2008. Flowering patterns in a seasonal tropical lowland forest in Western Amazonia. Biotropica. 40, 559567. Stickler, C.M., Nepstad, D.C., Coe, M.T., McGrath, D.G., Rodrigues, H.O., Walker W.S., Soares Filho, B.S., Davidson, E.A 2009. The potential ecological costs and cobenefits of REDD: a critical review and case study from the Amazon region. Global Change Biology 15: 28032824. Svenning, J C., Engelbrecht, B.M.J., Kinner, D.A., Kursar, T.A., Stallard, R.F., Wright, S.J., 2006. The relative roles of environment, history and local dispersal in controlling the distributions of common tree and shrub species in a tropical forest landscape, Panama. Journal of Tropical Ecology. 22, 575586. Ti cktin, T. 2004. The ecological implications of harvesting nontimber forest products. Journal of Applied Ecology. 41: 1121. Tonini, H., Kaminski, P.E., Costa, P.da., Schwengber, L.A.M., 2008. Estrutura populacional e produo de Castanhadobrasil ( Bertholletia excelsa Bonpl.) e Andiroba ( Carapa sp.) no sul do Estado de Ror aima. In: Wadt, L.H.O. (Ed.) Anais do 1 No madeireiros na Amaznia, Rio Branco, Acre. Tuljapurkar, S.D. 1982. Population dynamics in variable environments. III. Evolutionary dynamic s of r selection. Theoretical Population Biology 21:141165. Tuljapurkar, S.D. 1990. Population dynamics in variable environments. Springer Verlag, New York, New York. Valle, D., Phillips, P., Vidal, E., Schulze, M., Grogan, J., Sales, M., van Gardingen, P 2007. Adaptation of a spatially explicit individual treebased growth and yield model and long term comparison between reducedimpact and conventional logging in eastern Amazonia, Brazil. Forest Ecology and Management 243: 187198. Van Schaik, C.P., Terb orgh, J.W., Wright, S.J., 1993. The phenology of tropical forests: adaptive significance and the consequences for primary consumers. Annual Review of Ecology and Systematics. 24, 353377. Vander Wall, S.B., 2001. The evolutionary ecology of nut dispersal. Botanical Review. 67, 74117. Vander Wall, S.B., Kuhn, K.M., Beck, M.J., 2005. Seed removal, seed predation, and secondary dispersal. Ecology. 86, 801806.

PAGE 138

138 Varghese, A., Ticktin, T. 2008. Regional variation in nontimber forest product harvest strategies, trade, and ecological impacts: the case of black dammar ( Canarium strictum Roxb.) Use and conservation in the Nilgiri Biosphere Reserve, India. Ecology and Society 13: 11. [online] URL: http://www.ecologyandsociety.org/vol13/iss2/art11/ Verissimo, A., Barre to, P., Mattos, M., Tarifa, R., Uhl, C. 1992. Logging impacts and prospects for sustainable forest use in an old Amazonian forest frontier: the case of Paragominas. Forest Ecology and Management 55: 169199. Vieira, S., Trumbore, S., Camargo, P.B., Selhors t, D., Chambers, J.Q., Higuchi, N., Martinelli, L.A., 2005. Slow growth rates of Amazonian trees: Consequences for carbon cycling. Proceedings of the National Academy of Sciences. 102, 1850218507. Vincent, R., Binkley, C.S., 1993. Effiecient multiple use forestry may require landuse specialization. Land Economics 69: 370376. Vogt, W.P. 2005. Dictionary of Statistics and Methodology: A Nontechnical Guide for the Social Sciences. Sage Publications Inc., London. 355p. Wadt, L.H.O., Kainer, K.A., Gomes Silva D.A.P., 2005. Population structure and nut yield of a Bertholletia excelsa stand in Southwestern Amazonia. Forest Ecology and Management. 211, 371384. Wadt, L.H.O., Kainer, K.A., Staudhammer, C.L., Serrano R.O.P. 2007. Sustainable forest use in Brazili an extractive reserves: Natural regeneration of Brazil nut in exploited populations. Biological Conservation 141 : 332346. Wagner, H.H., Fortin, M.J. 2005. Spatial analysis of landscapes: concepts and statistics. Ecology 86: 19751987. Walker, R. 2003. Mapping Process to Pattern in the Landscape Change of the Amazonian Frontier. Annals of the Association of American Geographers 93 : 376398. Waller, D.M., 1979. Models of mast fruiting in trees. Journal of Theoretical Biology. 80, 223232. Webb, C.O., Peart, D.R., 2000. Habitat associations of trees and seedlings in a Bornean rain forest. Journal of Ecology. 88, 464478. Whitmore, T.C., 1984. Tropical Rain Forests of the Far East. 2nd edn. Oxford University Press, Oxford. Whitmore, T.C. 1990. An Introduction t o tropical rain forest. Clarendon, Oxford, England.

PAGE 139

139 Whitmore, T.C., Sayer J.A 1992. Tropical deforestation and species extinction. Chapman and Hall. London, UK. Wright, S.J., Carrasco, C., Caldern, O., Paton, S., 1999. The El Nio Southern Oscillation, variable fruit production, and famine in a tropical forest. Ecology. 80, 16321647. Wright, S.J., Jaramillo, M.A., Pavon, J., Condit, R., Hubbell, S., Foster, R., 2005 Reproductive size thresholds in tropical trees: variation among individuals, species and forests. Journal of Tropical Ecology, 21, 307315. Zarin, D.J., Schulze, M.D. Vidal, E., Lentini M 2007. Beyond reaping the first harvest: management objectives for timber production in the Brazilian Amazon. Conservation Biology 21: 916925. Zhang, Y. 2005. Multipleuse forestry vs. forestlanduse specialization revisted. Forest Policy and Economics 7: 143156. Zimmerman, J.K., Wright, S.J., Calderon, O., Pagan, M.A., Paton, S., 2007. Flowering and fruiting phenologies of seasonal and aseasonal Neotropical forests: The role of annual changes in irradiance. Journal of Tropical Ecology. 23, 231251. Zuidema, P.A., Boot, R.G.A., 2002. Demography of the Brazil nut tree ( Bertholletia excelsa ) in the Bolivian Amazon: impact of seed extraction on recruitment a nd population dynamics. Journal of Tropical Ecology. 18, 131. Zuidema, P.A., Jongejans, E., Chien, P.D., During, H.J., Schieving, F. 2010. Integral projection models for trees: a new parameterization method and a validation of model output. Journal of Ecology 98: 345355.

PAGE 140

140 BIOGRAPHICAL SKETCH During and after completion of her undergraduate degree, Christie Klimas research focused on the role of forests in global climate change and the carbon cycle. In North Carolina, she worked for Brookhaven National Lab at the Free Air Carbon dioxide Exchange (FACE) site. The FACE system allows for CO2 enhanc ement of entire forest stands. Klimas examined the effects of rising atmospheric carbon dioxide levels coupled with seasonal temperature increase on plant photos ynthetic carbon uptake. This work led her to the Biosphere 2 in Arizona where she studied plant carbon uptake in a more controlled environment. She learned a key component of climate change research: forests play a crucial role in the carbon cycle, and it is hoped that they will help mitigate the effects of increasing atmospheric CO2. However, the benefits that forests provide are being compromised by rapid and unprecedented changes in forest use and abuse. To learn more about the human dimensions of envir o nmental change, Klimas spent eleven months in Acre, Brazil as a Rotary Ambassadorial Scholar. She collaborated with rubber tapping communities to inventory reproducing adults of two species of rainforest trees Carapa guianensis (andiroba) and Euterpe prec atoria (aa), which have broad potential markets in various industries from food to medical. This investigation was part of a larger research initiative, which focuses on economic incentives that may encourage communities to seek sustainable use of the forest. Klimas had the privilege of joining other researchers, university faculty, government officials and motivated individuals in the pursuit of solutions to local problems that collectively compose some of the greatest existing scientific challenges. Co ncurrently frustrating, euphoric, rewarding and challenging, it was her first experience with applied science. As with all of her research endeavors, the more she learned, the more she

PAGE 141

141 realized she still needed to learn. It was the desire to improve her research efforts that led her to the interdisciplinary masters program at the University of Florida (UF). Her experience with the masters program at UF has strengthened her desire to continue collaborative research on using the earths resources in a w ay that balances ecological concerns with a respect and improved quality of live for resource users. Klimas hopes to remain in academia as a professor using science, interdisciplinary knowledge and her collaborative skills to mentor others. Research is an exciting, challenging and constantly changing field and she plans to use the skills and knowledge acquired through community based work in Brazil, research experience and Floridas interdisciplinary program to apply rigorous science to help solve/clarify s ome of the most intractable problems faced in tropical regions Through dis semination of my results, she hopes to contribute to economically sound, socially equitable development in the Amazon and throughout the globe. Klimas received her PhD from the University of Florida in 2010. She is currently teaching at DePaul University in Chicago, Illinois and resides with her husband and daughter.